Had the headline read instead, "Had an Algorithm Written This, Would You Have Realized It?", using the correct and elegant clause in the subjunctive mood, followed by a clause in the potential mood, then I would have doubtlessly assumed that a properly programmed algorithm had written it, rather than the improperly programmed Millennial intern who seems to write everything these days.
4
"What does it mean to be human?"
The difference between a robo-writer and a human writer is that the human understands what he/she is writing about, or for that matter what writing is. Just because a computer can produce the text "baseball game" doesn't mean it has any concept of what baseball is or why one would want to write about it.
Think about it this way--what if a person were to manually work through a program's writing algorithm to produce text, just as the program does? Would this process be what we call "writing"? What these programs do and what an author does are two fundamentally different kinds of activities--the former rote and mechanical and the latter creative and thoughtful. The economy of the future will prize the latter set of qualities highly, as computers are not even close to being able to emulate them.
The difference between a robo-writer and a human writer is that the human understands what he/she is writing about, or for that matter what writing is. Just because a computer can produce the text "baseball game" doesn't mean it has any concept of what baseball is or why one would want to write about it.
Think about it this way--what if a person were to manually work through a program's writing algorithm to produce text, just as the program does? Would this process be what we call "writing"? What these programs do and what an author does are two fundamentally different kinds of activities--the former rote and mechanical and the latter creative and thoughtful. The economy of the future will prize the latter set of qualities highly, as computers are not even close to being able to emulate them.
We can look forward to the day when ALL writing--and all reading--will be done by computers. Human beings are becoming obsolete, and I'm afraid we won't be missed.
3
Yes, but can a machine convincing tell us that it was at an event when it wasn't, the way Brian Williams and Bill O'Reilly can?
I don't think so...
I don't think so...
1
I have an algorithm which reads this stuff and gives me an executive summary.
3
I will admit that this dystopian development that combines elements from "1984" and "Brave New World" snuck up on me. The implications are both obvious and frightening. As all of our major corporate institutions continue to grow, consolidate, and merge with one another, the problems of information control become paramount. We already look quaintly at the world of "Mad Men" and their often clumsy and obvious efforts to manipulate human behavior, but this new generation of algorithms appears to be sufficiently sophisticated that most people won't even perceive what is coming. Public opinion, especially political opinion, is already manipulated by a coterie of didactic talking heads but at least their mistakes, misstatements, and factual errors can easily be spotted and checked. What will happen when the "talking heads" become droids whose mellifluous voices are linked to the Internet via impossibly complex algorithms that no human mind can keep up with? I would be willing to bet that Rupert Murdoch has already invested heavily in the research and development of the first "News Bot."
3
"Down, down in the basement, we hear the sound of machines." (Hat tip, and LOL, Talking Heads, circa 1983.)
3
Your test is silly. Mostly headlines. So what if they are written by computer.
"These software processes, which are, after all, a black box to us, might skew to some predicated norm, or contain biases that we can’t possibly discern. "
So in other words, just like human writers.
So in other words, just like human writers.
This feels like the self checkout to me or ATMs that can easily deposit checks. Another way that a very useful technology devalues human capital and ultimately takes away job opportunities.
Reading this story, I can't help but recall the movie Wall-E where humans were over weight and docile. They no longer did anything for themselves and instead relied on robots and automation. Guess what. We are already heading there.
What does it mean to be human? I think we are still searching for that definition in the 21st Century. Maybe an algorithm will tell us.
What does it mean to be human? I think we are still searching for that definition in the 21st Century. Maybe an algorithm will tell us.
There is a huge difference between form and substance. Algorithms can be reasonably good at reproducing form (there is really nothing to the "baseball" story except the scores and events, and the text context is just standard filler).
But let us ask it a question and see how it does.
In any case, companies can only get away with this nonsense because so many human writers these days are producing little but strings of vacuous cliches.
It doesn't matter whether the piece is produced by an electronic or a human robot reporter, an intelligent reader is going to ignore it. Anyone else deserves what they get.
But let us ask it a question and see how it does.
In any case, companies can only get away with this nonsense because so many human writers these days are producing little but strings of vacuous cliches.
It doesn't matter whether the piece is produced by an electronic or a human robot reporter, an intelligent reader is going to ignore it. Anyone else deserves what they get.
1
I'm constantly irritated, and still baffled, by how quickly and easily human beings have granted so much credit to computers for intelligence. ALL the intelligence, ALL the knowledge, ALL the decision-making ability is the result of whatever has been built into the software programming that is executed. Whatever a computer does is a display of human input. The big deal with a computer is that it is done very quickly, and repeatably. When the computer is connected to any kind of sensors, some very quick and precise decisions can result, but the decisions have been made beforehand and are all waiting in the programmed code for a particular situation to trigger their presentation. Computers don't have the 'magic' that seems routinely attributed to them. They are machines with stored answers and responses, which are the result of thought applied ahead of time. It's just that simple.
3
Humans create human-like things so that they can be more than humans, only, they become nothing.
It is all well and good for the author to claim that something was computer generated, but nothing was said about how the computer obtained the data that was contained in the two samples and that there was significant human interaction at the outset. I have serious doubts that an algorithm cooked up those paragraphs from scratch and the premise of the article is fairly disingenuous.
I have wondered about the articles I have been reading lately. So many seem to miss having a point or trail of into inconsequential nothingness.
Inasmuch as real people wrote the software code, doesn't that make the computer's output amount to merely one degree of separation from human origin?
I always suspected that Larry Eisenberg was an automated limerick generator:)
2
Given that most humans think and write badly, what does it mean for a machine to think and write like a human?
1
Goebbels is sitting in his little hell thinking: Damn! I was born a century too soon.
The tools available today to control people are simply incredible. All one can hope for is our energy to run out so that we come back to our senses (literally and figuratively speaking).
Written by a human being
The tools available today to control people are simply incredible. All one can hope for is our energy to run out so that we come back to our senses (literally and figuratively speaking).
Written by a human being
1
What do you say when a candidate gets most of the popular vote, but still loses the election?
That's a sad algorithm!
That's a sad algorithm!
Cute examples, but they work mostly because the phenomenon is relatively new. Will the computers be able to write more than boiler plate language? Most of the examples are from highly conventionalized types of texts, which often made people write as if they were machines even before computers were invented. And Ronald Reagan apparently produced such texts as a sports reporter in his youth.
1
Another question one has to ask is how this will "rig" the markets as you just put the algorithm on the story, set your parameters on where you want the various focuses to be and off it goes. I have used the "free" Narrative Scients bot they offer that tells a story from your Google stats. It's boring as heck but has a place over all. Let's say you have a staff meeting coming up and you need a "story" format to share with owners, employees, etc., put the bot to work on that spreadsheet.
Now on the web, we are talking a different story as again news is what the stock bots read. Eric from Nanex and I, months ago were talking about this and how it can rig the news and then a few weeks ago MIT came out with an article kind of discussing the same thing, so long and short of all of this, "if there's a way to rig something with code, people will do it"...nobody sees it and it runs in the background.
We all should know by now that us as humans are secondary readers of the news, the bots get it first during trading hours of course. Nanex is really good about the SIP feed and time markings so you can see when the bot news hits. When you see action "before" the bot hit, well then we have another discussion on who's getting the news ahead of time and that kind of rigging does occur.
Now on the web, we are talking a different story as again news is what the stock bots read. Eric from Nanex and I, months ago were talking about this and how it can rig the news and then a few weeks ago MIT came out with an article kind of discussing the same thing, so long and short of all of this, "if there's a way to rig something with code, people will do it"...nobody sees it and it runs in the background.
We all should know by now that us as humans are secondary readers of the news, the bots get it first during trading hours of course. Nanex is really good about the SIP feed and time markings so you can see when the bot news hits. When you see action "before" the bot hit, well then we have another discussion on who's getting the news ahead of time and that kind of rigging does occur.
I hope the New York Times will adopt journalistic standards for computer generated stories. If a human reporter would be credited by name, the program should be credited in some fashion. I'm not sure what readers need to know. Maybe something like this: "Today the newsroom welcomes Auto-Intern Version 6.2 to our staff. It will be drafting obituaries for the not-yet-dead."
4
Before making grandiose hypotheses, wouldn't one have to know exactly what data was given to the computer to write that first sports paragraph? The article never answers the obvious questions that a human would ask:
What data was given? That is, was it some combination of the box score, the play by play, the outcome of every pitch, or something even more detailed, such as a recording of the entire game by the play-by-play announcers? And while the computer may have known that, for example, the relief pitcher was the fourth left-hander since Sandy Koufax to have reached a certain number of strikeouts in a given number of innings, would it think that fact was newsworthy enough to include?
The computer didn't just "know" anything at all about the game. Some human fed it information, but in what form? Since the article ignores the desire of real people to want some knowledge about the process, and blithely runs on with its conclusion, that's a good sign that it was written by a computer. My conclusion: "Shelley Podolny" is a computer.
What data was given? That is, was it some combination of the box score, the play by play, the outcome of every pitch, or something even more detailed, such as a recording of the entire game by the play-by-play announcers? And while the computer may have known that, for example, the relief pitcher was the fourth left-hander since Sandy Koufax to have reached a certain number of strikeouts in a given number of innings, would it think that fact was newsworthy enough to include?
The computer didn't just "know" anything at all about the game. Some human fed it information, but in what form? Since the article ignores the desire of real people to want some knowledge about the process, and blithely runs on with its conclusion, that's a good sign that it was written by a computer. My conclusion: "Shelley Podolny" is a computer.
1
The Machine never freed us to pursue noble pursuits. Because we need more time, to make more money to meet the demands - both forced and self created.
Also, there is no such thing as noble task and mundane. The Joy is in the doing of little things not big success/destination but in journey. The world wide philosophy of making it big and then retiring to a paradise has corrupted us. I need work, EVERY DAY, like I need food everyday.
I have been writing little poems and the joy of writing each one lingers on for couple of weeks. So far, I have not published them. I don't know if they have linguistic quality and value. But I cherish the moments of first creative spark.
Writing is not for an end but for the Joy of writing. So, are we going to give up the THERAPEUTIC value of it, like we have given up so many handy crafts, to machine?
Also, there is no such thing as noble task and mundane. The Joy is in the doing of little things not big success/destination but in journey. The world wide philosophy of making it big and then retiring to a paradise has corrupted us. I need work, EVERY DAY, like I need food everyday.
I have been writing little poems and the joy of writing each one lingers on for couple of weeks. So far, I have not published them. I don't know if they have linguistic quality and value. But I cherish the moments of first creative spark.
Writing is not for an end but for the Joy of writing. So, are we going to give up the THERAPEUTIC value of it, like we have given up so many handy crafts, to machine?
2
Robo-journalism could only work, if at all, for a short statistics-laden news account, reporting scores or financial (breaking) news, again no more than a paragraph or two, about, say, the most active stocks traded. For any other writing, particularly novels and poetry or complicated science articles, the idea of this is chilling and also ridiculous. It would be the end of nuanced writing. How, for example, is a robo-writer to distill material to fit a word count, which requires judgment and balance. Above all, writing is so much about process and as well choices made by the author or journalist. As well, the essence of writing, to my thinking, is re-writing. Robo-journalism equals hack journalism performed by a dead wood reporter.
If money is speech and corporations are people, why shouldn't algorithms have an independent voice and be people too?
Equal rights and self-determination for algorithms!
Equal rights and self-determination for algorithms!
3
No surprise here. I first read of these in Orwell's *1984*, where he called them "versificators".
Science Fiction wins again.
Science Fiction wins again.
1
I took a look at the "Patient's Guide...." book. It is nothing but a badly done cut and paste plagarism of DHHS documents that are in the public domain. This hardly qualifies as a "technical book" in any meaningfull sense.
Seems like its very easy to write a fear mongering article about anything done by a machine and get it published.
1
Over the decades, it's been a fairly reliable science fiction theme to depict advanced societies that have prohibited the use of computers for complex tasks such as the journalism that this op-ed describes. Obviously, the authors were better at predicting the arc of innovation to visualize what might occur (and now has)
Clearly, the message is that some things, among them communication, must be reserved for carbon-based life forms, not silicon-based analogs.
Clearly, the message is that some things, among them communication, must be reserved for carbon-based life forms, not silicon-based analogs.
I guess I just have to get my robots to read all the stuff for me.
"No one believes it is happening now." The Singularity, that is.
How can we say it's not human writing, if the computer is programmed by humans?
Or, how can we say that most "human" journalistic writing is not also programmed, by the audience -- considering that it's written to very defined tastes, expectations and rules of a target readership?
Or, how can we say that most "human" journalistic writing is not also programmed, by the audience -- considering that it's written to very defined tastes, expectations and rules of a target readership?
Interesting because I noticed recently that I read 2 separate stories that didn't make sense. I reread them and tried to pluck coherence from the nonsensical to no avail. I thought that's odd. It was not a style issue. It was the words put together communicated nothing. So I asked a sibling and he recounted similar experiences. Thank you. Now I know to just stop when I'm getting nothing and abandon this computer generated garbage.
There is a very well known science?fiction concept that can be summarized as follows:
All advanced organic ie human societies are eventually taken over by their AI progeny and the humans destroyed. As the AIs are practical beings without much curiosity, they never expand beyond their own solar systems, explaining the apparent emptiness of the universe vis a vis intelligent life. The universe is actually filled with "mechanical life", but organic life fights a never-ending and inevitably futile war against extinction.
Variation on a theme stolen from Gregory Benford, but makes sense to me. And if I,m wrong, the Singularity is going to squash us anyway.
All advanced organic ie human societies are eventually taken over by their AI progeny and the humans destroyed. As the AIs are practical beings without much curiosity, they never expand beyond their own solar systems, explaining the apparent emptiness of the universe vis a vis intelligent life. The universe is actually filled with "mechanical life", but organic life fights a never-ending and inevitably futile war against extinction.
Variation on a theme stolen from Gregory Benford, but makes sense to me. And if I,m wrong, the Singularity is going to squash us anyway.
How different is developing an algorithm to write content from developing a "style" with which to write content, then writing it all yourself?
In both cases, a human identifies a pattern in subject matter (events, people, etc.), develops a format, and applies appropriate grammar and syntax rules. In the former case, technology carries out the task faster than any human could; in the latter, a human (who could be improving, varying, or updating the algorithm) spends the time to write.
Not all writing can be outsourced to computers yet, but the writing that benefits should be. (Replace "writing" with most gerunds and the condition holds.)
In both cases, a human identifies a pattern in subject matter (events, people, etc.), develops a format, and applies appropriate grammar and syntax rules. In the former case, technology carries out the task faster than any human could; in the latter, a human (who could be improving, varying, or updating the algorithm) spends the time to write.
Not all writing can be outsourced to computers yet, but the writing that benefits should be. (Replace "writing" with most gerunds and the condition holds.)
...and yet Google cannot make a coherent automated Japanese-to-English translation. Unless Google Translate is not using state-of-the-art algorithms, this fact alone leads me to believe that we have a long way to go.
Natural language generation is starting to revolutionize data analysis and reporting within enterprises, and is likely to become the preferred output, beating dashboards, graphs and charts and the like, and integrating them into narratives, just like humans.
More than 30 years ago, when I was an undergraduate computer science student, a friend and I wrote a program to generate Penthouse "Forum" stories. There was no pretense to artificial intelligence here, as the style and vocabulary of this writing was so formulaic that it easily lent itself to a rather simple script-driven approach and a smallish vocabulary of "technical" terms. It succeeded in creating laughable but almost plausible parodies of the original. But then, the original was practically a parody to begin with, enough to make us wonder whether the publisher was using an algorithm similar to ours.
There are many other types of writing that sufficiently formulaic that they inherently lend themselves to algorithmic generation. Given the "brute force" of computing power now available at minimal cost, it should not be surprising that software is taking over writing that formerly would have required a human with some amount of creativity. Given the low level of literacy so many people have today, the generated product doesn't even have to be very good. If it's news reporting, it only has to report the facts, possibly with the specific slant the publisher wants to give it (think Rupert Murdoch, for example).
Besides, the more human jobs are replaced with computers, the more shareholder value the executives who lay off the humans can create! That's what matters, after all.
There are many other types of writing that sufficiently formulaic that they inherently lend themselves to algorithmic generation. Given the "brute force" of computing power now available at minimal cost, it should not be surprising that software is taking over writing that formerly would have required a human with some amount of creativity. Given the low level of literacy so many people have today, the generated product doesn't even have to be very good. If it's news reporting, it only has to report the facts, possibly with the specific slant the publisher wants to give it (think Rupert Murdoch, for example).
Besides, the more human jobs are replaced with computers, the more shareholder value the executives who lay off the humans can create! That's what matters, after all.
1
Isn't it ironic that computers may already be better at distinguishing a human-written piece from a computer-written piece...
A slightly modified Blockhead Machine could (actually, it would have to, eventually) write Joyce's "Ulysses" or Tolstoy's "War and Peace." The difficultly would be in determining everything that isn't "Ulysses" or "War and Peace." Realize that when you are presented with a piece of machine generated prose or poetry that seems like human writing, what you are really seeing one of the "hits" and none of the "misses." And as with all magic tricks, I guarantee that it was a human who picked that "hit" out for you to see.
1
Two points:
1) For every algorithm that generates natural language, there's an analogue algorithm to detect whether it was generated by a human or machine. Once there are no longer algorithms that can do so, and humans can't tell them apart, then we've really created a machine as intelligent as a human, so we should respect it as such.
2) Let's not forget that human journalism can also be colossally bad. Recent events remind us that even mainstream reputable news organizations fabricate events, plagiarize, and draw biased conclusions based on poor or incomplete sources. Wouldn't then a machine be somewhat refreshing? Isn't it the paragon of objectivity we often strive for?
1) For every algorithm that generates natural language, there's an analogue algorithm to detect whether it was generated by a human or machine. Once there are no longer algorithms that can do so, and humans can't tell them apart, then we've really created a machine as intelligent as a human, so we should respect it as such.
2) Let's not forget that human journalism can also be colossally bad. Recent events remind us that even mainstream reputable news organizations fabricate events, plagiarize, and draw biased conclusions based on poor or incomplete sources. Wouldn't then a machine be somewhat refreshing? Isn't it the paragon of objectivity we often strive for?
5
I read news articles for 1.) the facts and 2.) insight. I guess an alogrithm will eventually be able to extract something akin to insight from its data set. And what will humanity have gained? Little I think.
1
Other examples of text composed by algorithms include legal documents such as wills, powers of attorney, health proxies, etc., and genealogical reports spewed out by software such as FamilyTreeMaker.
Not to panic. It's like playing chess against a computer program opponent. We can still out-think it and it remains to be seen how well a computer algorithm can generate new gestalts. In any event, of more pressing concerning is avoiding the fate of Emilio Estevez in the Bishop of Battle.
1
Well these sort of algorithms work because most writing is trite and hackneyed, especially sports stories (as quoted in the article). We do not read a sports story for human feelings or in depth thought--we only want the data, i.e., who won, who lost, by how much. Same goes for many novels and poetry, especially the modern stuff--which makes little sense even if written by a human. In the end, it really shows that, because of our lack of critical thinking/reading, we let a lot fly by us.
7
Uhh, sorry, it was obvious that a computer wrote the first passage. Why? The use of cliche. "Things looked bleak" ... snore. "Key single" ... a what single?? Pretty good for a machine, I must say. But get past formulaic and then I'll believe machines have joined the conversation.
1
Writer's block solved!!!! (but you apparently still need the facts!!! which can be much more troublesome to acquire than the form.... (So far as translating --- yes context is all when words have multiple meanings.... I was thinking about how to translate day-off into French which is not the same as off-day, e.g,)
Can you tell which one of these two sentences was written by a computer?
Sentence 1: My name is Jon.
Sentence 2: May name is Claire.
The fact that you cannot, implies that computer are as good as humans at writing text. Q.E.D.
Sentence 1: My name is Jon.
Sentence 2: May name is Claire.
The fact that you cannot, implies that computer are as good as humans at writing text. Q.E.D.
1
Look, I honestly don't mean to be smug or arrogant or the like, but I took the test of 8 questions, plus the 2 questions embedded in the article, and I was able to get the correct answer 9 out of 10 times. I don't think it's a big deal, because 10 questions is not that large of a sample, and perhaps these examples were not the most difficult ones that could have been generated.
What I think is noteworthy, though, is that there ARE differences between human processes and computerized processes which are discernible to the patient and "sensitive" or "experienced" mind. The issues raised in the article are extremely germane to our future on -- what is coming to be -- a more and more "automatized" planet. Thank you for this article, and I'll be on the look-out for more along these lines.
What I think is noteworthy, though, is that there ARE differences between human processes and computerized processes which are discernible to the patient and "sensitive" or "experienced" mind. The issues raised in the article are extremely germane to our future on -- what is coming to be -- a more and more "automatized" planet. Thank you for this article, and I'll be on the look-out for more along these lines.
Obviously written by a human - and a Western human, at that.
Self-centered.
Your problem, Shelley, is not when the computer replaces you as a writer.
It's when it replaces you as a reader.
Self-centered.
Your problem, Shelley, is not when the computer replaces you as a writer.
It's when it replaces you as a reader.
2
Let me get this straight: Plaigiarism is legal now?
1
Writing just a a human would is setting the bar too low. Sadly, most of the real humans I deal with write so poorly that an algorithm would certainly represent an improvement.
2
Spoiler alert....
One of the quiz items says 'The Russian novel "True Love" was written by a computer in St. Petersburg in 72 hours.' And was it written in Russian? If so, who wrote the English in the quiz? A human translator or a machine translation program? (Despite the proverbial "Never say never," I would bet that no algorithm will ever produce a completely adequate translation of literary prose of high quality.)
Although "convincing" formulaic reports on sports, weather, and financial markets can be generated by algorithms , it is usually possible for me to tell which news items in English at various websites have been written by literate native speakers, which by illiterate native speakers, which by non-native speakers, and which by computers. The reason is that errors of different and characteristic types are made by the latter three.
Here, for example, is a sentence I encountered at one of the many news aggregator sites to which Google News now displays links (you have to dig deep there before you find a legitimate news source): "Hoping to catch a glimpse of the next asteroid to zip previous the Earth?" A web search found, in a photo caption, the sentence that was changed by an algorithm to yield the above nonsense: "Hoping to catch a glimpse of the next asteroid to zip past the Earth?" This instance proves that an algorithm can be even dumber than a first-year undergrad who thinks that changing a few words will evade a charge of plagiarism.
One of the quiz items says 'The Russian novel "True Love" was written by a computer in St. Petersburg in 72 hours.' And was it written in Russian? If so, who wrote the English in the quiz? A human translator or a machine translation program? (Despite the proverbial "Never say never," I would bet that no algorithm will ever produce a completely adequate translation of literary prose of high quality.)
Although "convincing" formulaic reports on sports, weather, and financial markets can be generated by algorithms , it is usually possible for me to tell which news items in English at various websites have been written by literate native speakers, which by illiterate native speakers, which by non-native speakers, and which by computers. The reason is that errors of different and characteristic types are made by the latter three.
Here, for example, is a sentence I encountered at one of the many news aggregator sites to which Google News now displays links (you have to dig deep there before you find a legitimate news source): "Hoping to catch a glimpse of the next asteroid to zip previous the Earth?" A web search found, in a photo caption, the sentence that was changed by an algorithm to yield the above nonsense: "Hoping to catch a glimpse of the next asteroid to zip past the Earth?" This instance proves that an algorithm can be even dumber than a first-year undergrad who thinks that changing a few words will evade a charge of plagiarism.
1
Awesome: we already have algorithms that scour news reports for information; and now we have other algorithms that write the news reports in the first place. It's like a machine that bakes a cake and then eats it. Leaves the rest of us more time to watch TV.
Hopefully voting can also soon be replaced by an algorithm. I think Nate Silver's algorithms are getting pretty close to predicting the winner anyway, so why not just save everyone the trouble and let the algorithm do the actual electing?
In fairness though, Parker's book on Rosacea has a devastating review on Amazon, and I doubt that his other 999,999 books are much better.
Hopefully voting can also soon be replaced by an algorithm. I think Nate Silver's algorithms are getting pretty close to predicting the winner anyway, so why not just save everyone the trouble and let the algorithm do the actual electing?
In fairness though, Parker's book on Rosacea has a devastating review on Amazon, and I doubt that his other 999,999 books are much better.
15
We still invented it. The machine didn't.
Let's see a computer create a language and then write Romeo and Juliet.
Let's see a computer create a language and then write Romeo and Juliet.
2
Is it a coincidence that newswriters are often referred to as "hacks" and computer programmers are also referred to as "hackers"? There is a tremendous amount of routine and boring writing and boring computer programming. The more repetitious and predictable it is, the easier it is to automate.
1
No surprise here. I don't know about the details of the writing, but the Times uses a very limited number of liberal templates that are used to handle any news story. I would mention some, but unlike the 'bots, editors are touchy.
So ... this explains Ross Douthat's columns? Or Maureen Dowd's?
Just kidding.
Just kidding.
1
The fact that this article reports that several "news" organizations are using software to write stories and not using human reporting is quite disturbing and ethically unsound. I'll bet given this software and the lack of artistry required for Common Core Writing, these programs would probably get an A on every essay written.
1
If computers are able to persuasively imitate human decision-making in writing, then they also ought to be able to evaluate actual human writing. Computer-graded essays are a key component of the Common Core evaluations that American children will be taking next month, and this is disturbing for me, as a teacher and a citizen. I worry that my profession is being marginalized yet again, this time by technology, not politics, and that we as a species are accepting that computers can be programmed to mimic our minds. I want to believe that i?We are unique and unprogrammable, and I'm unwilling to budge on this.
Excuse me, but the computers can do this only because human being programmed them to do so. It is only formulaic manipulation of symbols. It is human in origin. It's a bad idea to let computers do this, but it's not the first time we've handed over what humans should be doing to computers. They are not writing content that is the equivalent of Shakespeare or any other great writer, nor shall they ever be able to do so. Not surprised the examples are sports writing! Algo-freaking-rithms can kiss my sweet derriere. And I am sure Mr. Turing would be appalled.
2
Hate to be pedantic, but really, writing isn't just about "writing". It's about reading. The value of what's written is in how it's received, not how it's generated. An automatically generated factual report delivers the same value as a manually created report. If you read the amount of ridiculous rubbish generated online, content value is the default editor of what's read and what's discarded. As a pro writer with about 8 million paid words of my own online, I can't say that I feel too threatened by algorithms which simply create shopping lists of organized language. Does an algorithm have a sense of humor? How about outrage, disbelief, derision, exploration, or new logical constructions, which are what human brains do when they're trying to work things out? Can an algorithm sneer or create nuances in multiple forms in the same sentence? I doubt it. Nice little pocket organizer, you've got there guys, but leave the real writing to us.
9
I managed to guess which sports story opening sentence was written by a human vs. the machine, because frankly the machine-written sentence did a better job of sticking to the basic who, what, where, when and how of the event being reported than did the human-written sentence, which included some minor extraneous information.
I was hoping that poetry would be more difficult for algorithms to compose, and so far it seems to be, but there are algorithms created to write poetry as well. And with the neo-surrealistic, non-sequiturish style, as well as the chopped-up prose style, of much contemporary American poetry, I suspect it will soon be more difficult to tell the difference between some human-written poetry in those styles and a machine-written poem in those styles.
I was hoping that poetry would be more difficult for algorithms to compose, and so far it seems to be, but there are algorithms created to write poetry as well. And with the neo-surrealistic, non-sequiturish style, as well as the chopped-up prose style, of much contemporary American poetry, I suspect it will soon be more difficult to tell the difference between some human-written poetry in those styles and a machine-written poem in those styles.
1
When the computers have taken over, would they have any reason to let us know?
(Or perhaps that question should be in the past tense.)
(Or perhaps that question should be in the past tense.)
1
Whoever or whatever wrote the lead sentences is a poor writer. Both sentences are run ons and lack journalistic flair. Does that mean we can no longer tell what good writing is?
1
Algorithms are designed by humans, not by software. It's algorithm-generated writing but human and subjective. The software may "learn" and refine itself but only at the direction of the coding humans.
It's designed for 2 things: speed and cost-efficiency. Don't expect it to be free of humans. It will have the fingerprint of the few hundred humans who create and greenlight it.
The problem? We're awash in an ever-expanding heap of duplicative, faux content that we mistake for real, for independent, for genuine.
It's designed for 2 things: speed and cost-efficiency. Don't expect it to be free of humans. It will have the fingerprint of the few hundred humans who create and greenlight it.
The problem? We're awash in an ever-expanding heap of duplicative, faux content that we mistake for real, for independent, for genuine.
8
Still written by a human because a human created the algorithm. It's not like the algorithm has free will. An algorithm is a nifty tool - not science fiction.
1
By this, I mean a poem like "Abide" by Jake dam York, published in the Times Sunday magazine, FEb. 22, 2015. It takes a soul to write a poem like that! Computers don't have souls.
2
The only reason you can't tell which blurb was written by an algorithm is that the human-generated one is so poorly written. Why would you need to indicate that Saturday was April 24? Who cares that Wilpon Baseball Complex is the home of "historic" Ray Fisher Stadium? My evaluation was that both blurbs were written either by an algorithm or by an inept human writer, and I stand by that assessment.
1
I believe the day is fast approaching when computers will be able to mimic human thought. But the reason won't be the power of computers, it's that our never ending interactions with machines and our unwillingness to challenge them makes us more stupid and less human by the day.
Our music is generated by computers, and there are auto tuned models "singing" or lip synching....our movies are generated on computers.....we play games that are nothing b
ut computer generated colors on a screen.......we communicate on screens.......our cars run using computers.......computers fly our airplanes and even direct the weapons that kill each other with us........we are reading this right now on a computer.....
ut computer generated colors on a screen.......we communicate on screens.......our cars run using computers.......computers fly our airplanes and even direct the weapons that kill each other with us........we are reading this right now on a computer.....
The written word defines us as a species and as an individual. It guides our culture and gives gravity our daily lives. I was a copywriter and was paid to make the leaps, twists or turns needed to create a winning ideas. I dare an advertising agency to replace their creative department with an algorithm and see how their clients react in time. Imagine being a literary agent today or an author? The robo-writer was incapable of writing the most important word in this editorial - salvage. We need to salvage the human equation from the hands of commerce that minimizes our all too human creative potential. First there was light; so we could read and write. Robo-writers need not apply . . .
It isn't he mechanization of industry that rips apart the creative heart and tears the profit motive out of the work of hard working writers. It's our industry heavies who see their employees as liabilities or worse. The $9.99 novel by Amazon and the $4.5 billion invested in comic books by a studio are but a symptom of what ails the human experience. Industry and government have a giant misunderstanding of what language, money and commerce can be, and we suffer unduly.
It isn't he mechanization of industry that rips apart the creative heart and tears the profit motive out of the work of hard working writers. It's our industry heavies who see their employees as liabilities or worse. The $9.99 novel by Amazon and the $4.5 billion invested in comic books by a studio are but a symptom of what ails the human experience. Industry and government have a giant misunderstanding of what language, money and commerce can be, and we suffer unduly.
9
It still takes humans to generate the algorithm. Computers only do what we humans tell them to do.
1
So computers can write poems. When they can write a poem with class, not hallmarky, that makes your skin produce goosebumps, makes you cry and tremble, and singes your mind, I'll be impressed.
2
Perhaps we could also pick our presidential candidates using an Al Gore rhythm?
1
Score seven out of eight on the quiz. Algorithm:
1. If the text contains numerical data, guess computer.
2. If the text is literary, do a web search.
1. If the text contains numerical data, guess computer.
2. If the text is literary, do a web search.
1
But alas, humans write the algorithms; even the algorithms that write the algorithms. We may be heading back to the age of the learned scribe with the rest of us enjoying life, free from the constraints of knowledge. It is a wearisome thing anyway, isn't it?
Now we need a program to read for us. But of course Google search does that for us. Where is the wire to plug into the back of my head?
1
All the examples here (sports writing, business reports) are very formulaic, so it's not surprising that an algorithm could do them. What an algorithm could NOT do is write good fiction, express and support an opinion (much like the piece right here), explain arcane facts in a clear manner. Until machines can do that, I'm not worried for writers.
I only missed one item on the quiz, but I knew the examples were either written by a person or generated by software. Had I not known, I would have assumed the quiz questions were written entirely by humans, some of them more talented than the others.
Took the quiz. Was easily able to separate human poetry and fiction from computer generated versions.
When it comes to reporting trivial data… it can go either way. There are only so many ways to say Team A play Team B today and team A won.
But An essay on, say, the evolution of Zone and Tampa 2 defense in football or a Runyon yarn about a washed up athlete would be a different story.
When it comes to reporting trivial data… it can go either way. There are only so many ways to say Team A play Team B today and team A won.
But An essay on, say, the evolution of Zone and Tampa 2 defense in football or a Runyon yarn about a washed up athlete would be a different story.
Given the fact that computers still cannot translate literary works of any complexity, perhaps human writers (and readers) are not on the brink of obsolescence.
1
The Turing Test is obsolete. Humans are not smart enough to tell the difference.
"If an Algorithm Wrote This, How Would You Even Know?"
(I hear the following in Rod Serling's voice.)
"Let me hazard a guess that you think a real person has written what you’re reading. Maybe you’re right. Maybe not."
For your consideration:
"Hillary Clinton’s record is more moderate than the Democratic primary voter today. So it was always likely that she would move left as the primary season approached. It’s now becoming clearer how she might do it. She might make a shift from what you might call human capital progressivism to redistributionist progressivism.
For many years, Democratic efforts to reduce inequality and lift middle-class wages were based on the theory that the key is to improve the skills of workers. Expand early education. Make college cheaper. Invest in worker training. Above all, increase the productivity of workers so they can compete.
But a growing number of populist progressives have been arguing that inequality is not mainly about education levels. They argue that trying to lift wages by improving skills is an “evasion.” It’s “whistling past the graveyard.”
The real problem, some of them say, is concentrated political power. The oligarchs have rigged the game so that workers get squeezed. Others say the problem is stagnation. It’s not that workers don’t have skills; the private economy isn’t generating jobs. Or it’s about corporate power. Without stronger unions shareholders reap all the gains."
So, what do you think?
(I hear the following in Rod Serling's voice.)
"Let me hazard a guess that you think a real person has written what you’re reading. Maybe you’re right. Maybe not."
For your consideration:
"Hillary Clinton’s record is more moderate than the Democratic primary voter today. So it was always likely that she would move left as the primary season approached. It’s now becoming clearer how she might do it. She might make a shift from what you might call human capital progressivism to redistributionist progressivism.
For many years, Democratic efforts to reduce inequality and lift middle-class wages were based on the theory that the key is to improve the skills of workers. Expand early education. Make college cheaper. Invest in worker training. Above all, increase the productivity of workers so they can compete.
But a growing number of populist progressives have been arguing that inequality is not mainly about education levels. They argue that trying to lift wages by improving skills is an “evasion.” It’s “whistling past the graveyard.”
The real problem, some of them say, is concentrated political power. The oligarchs have rigged the game so that workers get squeezed. Others say the problem is stagnation. It’s not that workers don’t have skills; the private economy isn’t generating jobs. Or it’s about corporate power. Without stronger unions shareholders reap all the gains."
So, what do you think?
I'm guessing an algorithm "wrote" this text. It's reads like it's the work of an automaton. Mass produced. Not modified by any externalities. Also kind of last-generation software. Apparently not self-correcting.
Scary. Of course we know robo-calls have become routine when trying to keep us on-line (your call is important to us, please stay on-line; all the while feeding us unsolicited commercials), or asking for donations (the initial call is a robo-call; and only when we answer, a human may pick it up), or giving us instructions or schedules for our intended flight or theater's movies. Well, you get the picture. But when we are fooled into thinking we are talking to a real person, and are not, it becomes potentially a fraud of sorts, unfair and even dangerous. Ideally, this must stop. Meanwhile, it pays to remain alert, and be all the more appreciative when a seductive, empathetic, and 'knowing' voice answers you; but even then, I wouldn't hold my breath. The robots are coming; and although of our own creation, as we feed them with the bottle at first, then with sophisticated programming far beyond beating us on chess, they may start acting on their own to yet unheard heights...and depths.
Similar to industrial automation, the question has become not "if" computers are capable of performing human activities, but "whether" they should.
Where will our self-worth reside when computers make our human talents useless? How will we live?
Why will we live?
Where will our self-worth reside when computers make our human talents useless? How will we live?
Why will we live?
2
Sorry pal, until there is actual-ai there ain't gonna be no
convincing "algorithmic" speech. It is not even clear
that speech acts can be completely defined by computational algorithms.
Nuance Corp notwithstanding , speech recognition systems don't
flipping work at flipping all ("I'm sorry I didnt get that, please say your
password again a bit louder so every body on the next floor can hear it...")
Furthermore probabilistic systems lke the jeopardy playing ibm system
are not "thinking machines", they are just big-data crunchers.
convincing "algorithmic" speech. It is not even clear
that speech acts can be completely defined by computational algorithms.
Nuance Corp notwithstanding , speech recognition systems don't
flipping work at flipping all ("I'm sorry I didnt get that, please say your
password again a bit louder so every body on the next floor can hear it...")
Furthermore probabilistic systems lke the jeopardy playing ibm system
are not "thinking machines", they are just big-data crunchers.
The examples shown are such short snippets of information, even the poetry ones, that they really don't discriminate much. While it is impressive that computer algorithms can put together comprehendible sentences containing snippets of information, that is not he same as actually composing and writing something substantial. For slightly lengthier and conceptually more developed paragraphs, I'll bet it would be easy to tell, even when the human is a mediocre (although perhaps not terrible) writer.
Manipulation is what makes us human.
Autowritten
New York Times headline asks "If an Algorithm Wrote This, How would You Even Know?"
Reading, an author responds "I would know." Another worries, was the article's author human, or an "automated narrative generator"?
Am I? Who, what, by the way, is writing this? Am I?
Confusion reigns, weather rains, words flow. Now which: machine?
New York Times headline asks "If an Algorithm Wrote This, How would You Even Know?"
Reading, an author responds "I would know." Another worries, was the article's author human, or an "automated narrative generator"?
Am I? Who, what, by the way, is writing this? Am I?
Confusion reigns, weather rains, words flow. Now which: machine?
1
..and how do you know that a human is reading this?
At some point, we'll all have algorithms to read a the stuff that algorithms are writing, to sum it up for quick consumption.
At some point, we'll all have algorithms to read a the stuff that algorithms are writing, to sum it up for quick consumption.
Teaching bad writing is already rife in school systems. The formulaic five paragraph essay is taught at every grade level. Algorithms might improve deadly dull and predictable writing...they can't do worse.
The algorithmically generated prize should go to whoever devises a tool that can digest the suffocating volume of output continually flung from every direction into a small enough relevant useful nugget to leave cranial space for reflection and inspiration.
While I suppose it is inevitable that machines will crank out more of what we read, I would like to see a practice whereby machine-written pieces, say news articles or online reviews, and labeled as such, so that we (still) human readers can decide how much credence they deserve.
.. and a computer algorithm is reading it.
3
Half way through my senior year in high school, back in 1978 in Sarasota, Florida, my mother kicked me out of the house for good. I hitchhiked up and down US 41 until Russ, who operated a body removal service that lugged dead bodies from highway accidents to the morgue, took me in.
In many ways, my new life was exciting and exotic. But, for all the excitement, the loneliness was often desperate and crushing. The body removal culture was profoundly lacking in teenagers.
Then one day I wandered out of Helen’s Used Bookstore with a copy of The Three Stigmata of Palmer Eldritch by Philip K. Dick . A month later, I left Helen’s with Albert Camus’ The Plague. After that, it was “The Seven Greatest Short Novels” by Tolstoy, Turgenev, Voltaire, Chekov, and three other authors I’ve since forgotten.
Honestly, in retrospect, my guess is that I barely understood any of them—except one thing. I understood that, if I were to sit down, night after night, curled up with the words of those great authors, I’d be able to struggle my way into college and a better life.
Just as importantly, my friendships weren’t unrequited. Even if those great writers hadn’t written their books for me personally, I was sure they had written their masterpieces for young men just like me. I was proud to offer them my imagination and loyalty.
I can’t imagine forging such deep and wonderful relationships with a series of impersonal circuits.
In many ways, my new life was exciting and exotic. But, for all the excitement, the loneliness was often desperate and crushing. The body removal culture was profoundly lacking in teenagers.
Then one day I wandered out of Helen’s Used Bookstore with a copy of The Three Stigmata of Palmer Eldritch by Philip K. Dick . A month later, I left Helen’s with Albert Camus’ The Plague. After that, it was “The Seven Greatest Short Novels” by Tolstoy, Turgenev, Voltaire, Chekov, and three other authors I’ve since forgotten.
Honestly, in retrospect, my guess is that I barely understood any of them—except one thing. I understood that, if I were to sit down, night after night, curled up with the words of those great authors, I’d be able to struggle my way into college and a better life.
Just as importantly, my friendships weren’t unrequited. Even if those great writers hadn’t written their books for me personally, I was sure they had written their masterpieces for young men just like me. I was proud to offer them my imagination and loyalty.
I can’t imagine forging such deep and wonderful relationships with a series of impersonal circuits.
2
I like the errors, nuance, self-doubt and nuance that come from real people.
The "virtual" world is a construct, and is not subject to the consequences that await us as we come to prefer pixels on a screen to the messiness of real life, and refuse to face the ethical questions that require our attention.
Makeup is surface, not content.
The "virtual" world is a construct, and is not subject to the consequences that await us as we come to prefer pixels on a screen to the messiness of real life, and refuse to face the ethical questions that require our attention.
Makeup is surface, not content.
9
The concept of robo-journalism is very disturbing to me, especially the prediction that 90% of news could be generated that way by the mid-2020's. Robots and algorithms are never going to investigate corruption like the NY Times and 60 Minutes and are never going to be able to create intelligent satire and parody like Jon Stewart and Stephen Colbert. We need human journalists to do those things.
If robo-journalism is accepted as mainstream, the likelihood that anyone will have the skills and training and funding to do what only humans can do is quite low. The Koch brothers, Rupert Murdoch, and other conservative fanatics are probably funding robo-journalism because they want investigative journalism, satire, and parody to die out. I hope I die before that happens...
If robo-journalism is accepted as mainstream, the likelihood that anyone will have the skills and training and funding to do what only humans can do is quite low. The Koch brothers, Rupert Murdoch, and other conservative fanatics are probably funding robo-journalism because they want investigative journalism, satire, and parody to die out. I hope I die before that happens...
30
Perhaps this is finally the time to separate "Reporting" from "Journalism". After all, it is humans who select and feed the data to the machines who "write it up".
1
More troubling than the fact the stories are generated by computer is the fact that they are not labeled as such. Such stories should be prefaced by the phrase "Text generated by computer, supervised by..." and then given the name of the person, who is ultimately responsible for the content.
It's not the stringing together of words that is remarkable, it is the lack of responsibility taken for their authorship.
It's not the stringing together of words that is remarkable, it is the lack of responsibility taken for their authorship.
56
Like GM foods.
1
The reason you can't tell the difference in authorship between the two sports passages is that neither has any nuance of emotion or thought or attitude, and neither is stylistically original or individualized. It's boilerplate.
Wake me up when a computer can write "Sailing to Byzantium"—or any piece of writing that's worth pondering, memorizing, and treasuring.
Wake me up when a computer can write "Sailing to Byzantium"—or any piece of writing that's worth pondering, memorizing, and treasuring.
54
You are so right! However, it is possible to tell that the second one was written by a person, because it is (somewhat) more idiosyncratic, using the less-often-used word "salvage," parentheses, and qualifying the name of the complex with a unique fact. The prior sample, by computer, contains cliched phrases, such as "Things looked" and "pull out."
Perhaps, eventually, programmers will compile a more interesting database of vocabulary and categorize subsets by various writing "personalities."
Perhaps, eventually, programmers will compile a more interesting database of vocabulary and categorize subsets by various writing "personalities."
8
This is called the 'super-human fallacy', an automaton is only significant if it can match or exceed the best human practitioners.
The counter argument is simple: wake me up when *you* can write "Sailing to Byzantium" --- or any piece that is worth memorizing and treasuring.
The counter argument is simple: wake me up when *you* can write "Sailing to Byzantium" --- or any piece that is worth memorizing and treasuring.
1
An insightful analysis. There is much agreement with CW'observations in the comments below.
1
It would be wonderfully ironic if the "NYT Picks" and the "Readers' Picks" were simply preselected by a program prior to publishing articles in the Times. For instance, any comment by Gemli would head the Readers' Picks but hardly ever appear in the NYT Picks. Also, anything well written by a conservative reader would qualify as an NYT Pick. As time passes the algorithm evolves into a highly sophisticated expert system, due to actual feedback from editors and readers alike. A simple genetic algorithm, if you will.
14
@ Richard I had no sooner sworn that I would not write another comment or reply today and then I visited Algorithm land where to my surprise I have two of the first comments, something the Times system failed to tell me, and I am enjoying reading some Verified comments that I do reply.
You state in your very short comment exactly what I have been writing to a Comment-Responsible Times Person. It appears to me and some of my Verified friends that the Times has tweaked the Select the Next Verified algorithm so that it will choose people who love David Brooks and who absoutely cannot write with the elegance of gemli or others I could name.
So when the Times Comment People tell me, no personal judgement is involved in picking Verifieds I take that with a grain of salt since some person had to formulate the algorithm.
RDeanB and CWolfe above you also have written my thoughts so I can close and write no more replies.
Only-NeverInSweden.blogspot.com
Never in Sweden will my Swedish newspaper DN have a decent (any) comment system! Aldrig!
You state in your very short comment exactly what I have been writing to a Comment-Responsible Times Person. It appears to me and some of my Verified friends that the Times has tweaked the Select the Next Verified algorithm so that it will choose people who love David Brooks and who absoutely cannot write with the elegance of gemli or others I could name.
So when the Times Comment People tell me, no personal judgement is involved in picking Verifieds I take that with a grain of salt since some person had to formulate the algorithm.
RDeanB and CWolfe above you also have written my thoughts so I can close and write no more replies.
Only-NeverInSweden.blogspot.com
Never in Sweden will my Swedish newspaper DN have a decent (any) comment system! Aldrig!
Quirk free and full of platitudes
And lacking in human attitudes,
Flaccid,, acquiescent,
Rarely percipient,
Writing by algorithmic dudes.
And lacking in human attitudes,
Flaccid,, acquiescent,
Rarely percipient,
Writing by algorithmic dudes.
22
"What does “human” even mean?" Having the money to pay for all this automation?
1
Exactly.
"It’s understandable. The multitude of digital avenues now available to us demand content with an appetite that human effort can no longer satisfy."
Come on. There are thousands of good, human writers out there, but companies don't want to pay them for what they're worth, so now, it's our human content appetite that's the problem. These programs can't protest or ask for raises. How dehumanizing for the reading public.
"It’s understandable. The multitude of digital avenues now available to us demand content with an appetite that human effort can no longer satisfy."
Come on. There are thousands of good, human writers out there, but companies don't want to pay them for what they're worth, so now, it's our human content appetite that's the problem. These programs can't protest or ask for raises. How dehumanizing for the reading public.
I'm a professional writer and I've been obsolete for years. Why? Because everybody thinks they are a writer these days.
47
You are only obsolete because you think you are. If you are truly a professional then "everybody" doesn't count.
Joe Pros,
They comes, they goes.
Poetry ain't easy,
So stick to the prose.
They comes, they goes.
Poetry ain't easy,
So stick to the prose.
2
Everybody thinks he/ she is an artist because it can doodle or graffiti something on a surface. Give me a break! anyone can write, it doesn't make them Shakespeare or Rembrandt.
1
Two questions. First question: If an algorithm writes a novel or article, who/what gets the copyright? Second question: Shouldn't there be someone / algorithm mulling over the need to legally inform readers that the ghost (writer) isn't in the machine, it is the machine?
3
Monkey theorem be damned! Long live the algorithm! The improbable becomes the probable, and humanity is left to wonder why. Alas, we have no one to blame but ourselves.
2
Soon, human being will no longer be needed to preserve humanity and may, in fact, become the greatest threat. Look at ISIS taking sledgehammers to 3,000 year old statuary. The question for those of us still alive is: What gets preserved? Our talents, insights and skills, we probably agree, should be codified into the digital species that succeeds us. But what about emotions, passion, belief. Will our digital heirs have religion that will lead to religious wars between machines or algorithms?
16
In the end, nothing gets preserved.
7
Or will our digital heirs be atheists leading to secular wars between machines or algorithms?
But some brides of quietness remain unravished longer than others.
What if a human reader could be fooled only some of the time? What if some people were fooled less often than others? What if the results of the Turing test were wildly inconsistent, all the time? ("You can't fool all the people all the time.") Then exactly what would this prove?
2
It's hard to not think of the formulaic nature of most news writing. Newspapers and magazines begin a story with a specific named human example, zoom out to consider the issue at large, quantify how controversial it is, and then give all sides equal weight without regard to merit.
"Myra Kaybaul has lived on this plot of land across the Seemaus River from Winden Poultry Processing for 35 years. A plain-spoken woman, she greets me in jeans, flannel and leather boots, the uniform of many a long-term resident here."
"A few years ago, she noticed her dogs were always restless, sometimes having trouble digesting their food. When she mentioned this to local rancher Sam Marsden, his laconic reply came right back: 'You and everyone else.'"
"Now many residents of Pilgree want answers, and their questions have created the first tension in the long-term relationship between residents of this quiet upstate Georgia town and its principal employer. Nearly every long-term resident has worked for Winden Poultry at some point, even if only for a summer or two during high school, but some have joined the newer inhabitants of the commuter tracts to the east of town in asking what Winden is dumping in the river and what the Pilgree City Council knows about it."
I am not a gifted programmer, but I could put together a program to write something like this. Aspects like descriptions of demeanor and clothing might reqire outside help building a catalog of relevant nouns and adjectives.
"Myra Kaybaul has lived on this plot of land across the Seemaus River from Winden Poultry Processing for 35 years. A plain-spoken woman, she greets me in jeans, flannel and leather boots, the uniform of many a long-term resident here."
"A few years ago, she noticed her dogs were always restless, sometimes having trouble digesting their food. When she mentioned this to local rancher Sam Marsden, his laconic reply came right back: 'You and everyone else.'"
"Now many residents of Pilgree want answers, and their questions have created the first tension in the long-term relationship between residents of this quiet upstate Georgia town and its principal employer. Nearly every long-term resident has worked for Winden Poultry at some point, even if only for a summer or two during high school, but some have joined the newer inhabitants of the commuter tracts to the east of town in asking what Winden is dumping in the river and what the Pilgree City Council knows about it."
I am not a gifted programmer, but I could put together a program to write something like this. Aspects like descriptions of demeanor and clothing might reqire outside help building a catalog of relevant nouns and adjectives.
7
Ouch. So poorly written, be it by algorithm or human.
5
As someone who writes both fiction and software... color me unimpressed. All the examples of computer "writing" in this article are simply regurgitating data. Certainly, someone thought up some clever tricks to make it read more naturally, but I see nothing to indicate that the computer is producing ideas or combining information in novel ways. Another poster compared it to Mad Libs, and that's exactly right.
When a computer produces a work of fiction that can't be told apart from something written by a skilled human, let me know.
When a computer produces a work of fiction that can't be told apart from something written by a skilled human, let me know.
23
My take-home lesson from the two writing snippets is that most contemporary human writing is of rather poor quality. Given that computers have become quite good at writing English, the only hope is for us humans to write better.
29
Not mentioned here is the growing reliance on computer algorithms to process job applications. But a job app filed by computer and read by another app is also going to treat people like we were apps ourselves. And if we don't fit neatly into the mold the program is designed to recognize, our skills and our experience, and our character, that is our human traits will begin to be ignored as irrelevant. There something dangerous and inherently contradictory about using an app to evaluate a person by ignoring the very things that differentiates people from programs. Sooner or later we're going to become irrelevant too.
26
"Just like a human would" means just like an AVERAGE human would. Algorithms are not conceptually original. They are constrained by their parameters and databases. This underscores the challenge to journalists or any other writer. Are you merely going through the motions? Or, are you a really good journalist and original writer?
10
Does the robo-writer algorithm also exist in Chinese or Japanese, for instance? Can these ideographs be interpreted in a similar manner to English to imnpart information?
1
First, algorithmic writing says more about modern human composition and information consumption than about technology. Second, sports and business information fall on the more discrete "facts" part of the information spectrum. I'll be more impressed when an algorithm writes a Summa Theologica and tells me something about God, or even about something greater than itself.
6
What interests me is the use of computers not to imitate human writers but to create texts only a computer could make. A human writer can only deal with a limited number of characters and plot lines. A computer could potentially produce a novel with a thousand coherent intersecting plots. It could produce a detective story so detailed and meticulous it would have no human correlate. Would such an item be readable? It would be a new form. If it were guided, overseen and shaped by a human intelligence, it could be remarkable.
4
Voiceofamerica, alas. But unremarkable
All machines would fail within few years without human care, as they cannot maintain themselves and have no interest in doing so. Likewise, the evolution and intention of machines depends upon human interests. And, if some of us decide our life is better off without machines, we can choose to abandon them and return to nature, if any of us remember how.
But, if Arnold arrives one day, nuclear Armageddon is on!
But, if Arnold arrives one day, nuclear Armageddon is on!
1
In the quiz an excerpt from the algorithmic book "True Love" is used. From an article about this book we have "...is the work of a computer program, created by a team of IT specialists and language experts." Aye, there's the rub.
This seems more the equivalent of a magic trick, based in illusion and subterfuge rather than reality. The means matters. A human writer is not assembled by a committee, and even more importantly has a body through which they have accumulated vast subjective experience which informs their writing. The computer approach seems closer to using modern technology to actually perform the million monkeys thought experiment. A million monkeys typing away might well produce something "Shakespeare-y" given 72 hours, especially if their output is continuously filtered by the encoded knowledge of human experts. And no, I'm not a Luddite terrified by the idea: I'm a programmer, and a writer.
This seems more the equivalent of a magic trick, based in illusion and subterfuge rather than reality. The means matters. A human writer is not assembled by a committee, and even more importantly has a body through which they have accumulated vast subjective experience which informs their writing. The computer approach seems closer to using modern technology to actually perform the million monkeys thought experiment. A million monkeys typing away might well produce something "Shakespeare-y" given 72 hours, especially if their output is continuously filtered by the encoded knowledge of human experts. And no, I'm not a Luddite terrified by the idea: I'm a programmer, and a writer.
4
The National Weather Service has used computer generated written forecasts for decades; however, the actual numbers in them are still produced by human forecasters. Even though the big computer simulations of the atmosphere produce sets of numbers, humans still beat the computers. The best philosophy seems to be let the computers do what they do best and people do what they do best.
5
"the actual numbers in them are still produced by human forecasters"
The US NWS point forecasts are computer generated:
http://www.srh.weather.gov/srh/jetstream/webweather/pinpoint_max.htm
The US NWS point forecasts are computer generated:
http://www.srh.weather.gov/srh/jetstream/webweather/pinpoint_max.htm
Big deal. Most news articles are the mental equivalent of breakfast cereals & snack foods & they've been manufactured following standard algorithms for years. It follows that big business considers what works for the body works for the brain.
4
readers of "1984" will remember with glee the description of books written by machines to be published for the proles.
gedanken experiment. if i input all the robotic media feeds into my algorithmic media scanning and product recommending computer, DVR and cellphone apps, then what am i? increasingly, just an output function of a robotic infrastructure. and what criteria determine the value of my output function for the infrastructure? wallet, above all else, then wealth that makes spending possible, and finally behavioral outcomes such as voting, attending, "using", "sharing" and so on. in other words, i become a resource, an incentive -- and a tool.
the insight that is missing in this story is not in the algorithm that can't quite yet develop a new interpretation of a shakespeare play -- it's in all of us, who scollectively till can't see the eventual culmination of a robotic infrastructure designed to exploit the profit potential of a continually growing human population. burn carbon, breed people, and use robotic media to distract the people from the fundamentals of their existence ... these are the premises of our era.
gedanken experiment. if i input all the robotic media feeds into my algorithmic media scanning and product recommending computer, DVR and cellphone apps, then what am i? increasingly, just an output function of a robotic infrastructure. and what criteria determine the value of my output function for the infrastructure? wallet, above all else, then wealth that makes spending possible, and finally behavioral outcomes such as voting, attending, "using", "sharing" and so on. in other words, i become a resource, an incentive -- and a tool.
the insight that is missing in this story is not in the algorithm that can't quite yet develop a new interpretation of a shakespeare play -- it's in all of us, who scollectively till can't see the eventual culmination of a robotic infrastructure designed to exploit the profit potential of a continually growing human population. burn carbon, breed people, and use robotic media to distract the people from the fundamentals of their existence ... these are the premises of our era.
3
"Imagination is more important than knowledge"said Einstein. Reporting information to the mass media has been done electronically for a long time. Making millions of bits of information tell you a story behind the numbers has been available since the dawn of the computer age. Perhaps the time will come when we realize that people are more important than machines and there is more value in that idea than the profit motive. Although technically it is possible to have a machine write copy, where does the story come from ? What is it's value to the culture or society from which it originates ? Box scores or justice ? There's the rub. Were's the money ? Computers cannot define an abstraction or elicit human emotion, nor can they create anything from scratch. Processing 0's and 1's has never created an idea, principle, set or moral values, or the belief in anything.
4
After every sentence in which you state what computer programs or robots cannot do, you should add the word "yet." The achilles heel of every new technology is power. When we finally burn all the oil and coal, there will not be enough energy to support the current number of people on this planet. Will we have robots or computers that can get by on solar, wind or geothermal power? Or will the remarkable ability of humans to survive on meager food sources trump the advanced technology? It would be truly ironic if we develop the ability to replicate ourselves in artificial form and are left so unimpressed that we decide to call the whole thing off. As history has shown, humanity does have a death wish.
Never mind humanity: my cat is conscious and comes up with new ideas, solves problems creatively, and algorithms do not. They can't replace my cat, even if cuddly little cat-robot bodies could be devised to house a cat-behaviour algorithm.
The old joke is that whether a computer can 'think' is as irrelevant as whether a submarine can 'swim' - the point is to get a certain job done, one that may not actually require consciousness, just a mechanism.
We think of only the body engaged in the work of shoveling when it is replaced by a backhoe, but shoveling actually requires vast brain activity to coordinate and balance the body and order dozens of muscles to work together.
So we think of accounting or engineering or writing as "brain work" and are worried by this new development, because now the machines have come for us white-colllar elites rather than Ned Ludd and his weavers. But only the mechanistic jobs of the brain are being replaced, the perfectly predictable functions.
"Humanity" is not at risk any more than from backhoes or cloth factories; just the literally sub-human parts of our jobs. The humiliating thing, I suppose, is the realization of how much of even desk jobs is actually beneath humanity's truly unique abilities.
The old joke is that whether a computer can 'think' is as irrelevant as whether a submarine can 'swim' - the point is to get a certain job done, one that may not actually require consciousness, just a mechanism.
We think of only the body engaged in the work of shoveling when it is replaced by a backhoe, but shoveling actually requires vast brain activity to coordinate and balance the body and order dozens of muscles to work together.
So we think of accounting or engineering or writing as "brain work" and are worried by this new development, because now the machines have come for us white-colllar elites rather than Ned Ludd and his weavers. But only the mechanistic jobs of the brain are being replaced, the perfectly predictable functions.
"Humanity" is not at risk any more than from backhoes or cloth factories; just the literally sub-human parts of our jobs. The humiliating thing, I suppose, is the realization of how much of even desk jobs is actually beneath humanity's truly unique abilities.
9
Algorithms don’t come up with lots of new ideas yet, but they will: http://spectrum.ieee.org/automaton/robotics/artificial-intelligence/face...
Hmm...I correctly identified the provenance of every quote in the quiz. I suspect that has a lot to do with being a teacher of literature. Critical, close reading of texts is a valuable skill for being able to identify various markers of authorial voice (or lack of it) and perspective which, though subtle, can convey important information about both the author and the writing's intended impact on the reader.
4
I suspect that it was at least partly luck.
No, you're the robot.
4
First, human beings labored under toil. Various kind of labor, everything from clearing a rocky pasture to crafting a careful paragraph to performing delicate surgeries to delivering pizza. The hot sun and tired mind and muscles makes work very hard, very often.
This toil bothered human beings. They gradually developed devices relieved them of their burdens. the devices were fruitful, and multiplied.
One day, the humans realized they had developed devices that made obsolete much need for many of them. Everyone made jokes about how the automobile replaced the horse-drawn carriage and the buggy whip, and everyone continued on apace.
Soon most people were replaced, except those who owned the expensive machines who did the work of one hundred humans for one hundred years and never complained or needed nothing more than a monthly repair inspection.
And that is where the story stops. The machine tried to suggest "And they lived happily ever after" as the final sentence, but you know how machines are: they only know how to write self-serving stories.
This toil bothered human beings. They gradually developed devices relieved them of their burdens. the devices were fruitful, and multiplied.
One day, the humans realized they had developed devices that made obsolete much need for many of them. Everyone made jokes about how the automobile replaced the horse-drawn carriage and the buggy whip, and everyone continued on apace.
Soon most people were replaced, except those who owned the expensive machines who did the work of one hundred humans for one hundred years and never complained or needed nothing more than a monthly repair inspection.
And that is where the story stops. The machine tried to suggest "And they lived happily ever after" as the final sentence, but you know how machines are: they only know how to write self-serving stories.
5
I often wonder media outlets like CNBC, Yahoo Finance, Fox News.. have robot talking heads, or humanoids reading machine generated scripts,. Hmmmmm..
1
I'm sure O'Reilly is a robot. No human could be that callous.
7
[[SundayReview | Opinion
If an Algorithm Wrote This, How Would You Even Know?]]
I think the more important question is, would we even care?
If an Algorithm Wrote This, How Would You Even Know?]]
I think the more important question is, would we even care?
1
Do the algorithms know not to split infinitives?
2
If they write in Latin, then yes. It's physically impossible to split an infinitive in Latin, but in English we can and do.
Sometimes we even - horrors! - truncate them. It's hard not to.
Sometimes we even - horrors! - truncate them. It's hard not to.
2
I just took the quiz and got 2 wrong out of the 8 texts. I attributed the two as being computer generated. I'm not sure what the lesson is. That some people write with the soul of an algorithm?
1
As for the video game side of it, what better game can there be than one where the avatar is organic/mortal and capable of evolving to the point where it is conscious of its mortality. Such an avatar would FEEL fear, would FEEL love, would FEEL desire ... actually, would BE us.
QED, WE are the video game creation of some bored teen immortal and the video interface is not just sight/sound/touch but a completely immersive experience where the immortal inhabits its creation and forgets its immortality.
Time to build some temples to our newly-discovered god.
QED, WE are the video game creation of some bored teen immortal and the video interface is not just sight/sound/touch but a completely immersive experience where the immortal inhabits its creation and forgets its immortality.
Time to build some temples to our newly-discovered god.
You may be onto something. There is a growing and well respected body of research that suggests we are living in a holographic universe and that our sense of intentionality is just an illusion.
And who makes the machines?
Too much careless talk of the "post-human." It's all human, in particular the least humane stuff.
Too much careless talk of the "post-human." It's all human, in particular the least humane stuff.
3
The only thing one needs to imitate sports writing is a sense of humor.
That being said, we aren't yet at the point where AI's can contribute human-level or higher intelligence to a book. They can merely mine it. That, of course, will change as processing power increases.
Unlike the author, I don't think that this will lessen us. We will still be what we are, only we will have more powerful thinking tools that will replace our limited intelligence just as power tools replaced Paul Bunyan's axe. Let us hope that we don't end up regretting the arrogance that we ourselves have shown to animals less intelligent than ourselves!
That being said, we aren't yet at the point where AI's can contribute human-level or higher intelligence to a book. They can merely mine it. That, of course, will change as processing power increases.
Unlike the author, I don't think that this will lessen us. We will still be what we are, only we will have more powerful thinking tools that will replace our limited intelligence just as power tools replaced Paul Bunyan's axe. Let us hope that we don't end up regretting the arrogance that we ourselves have shown to animals less intelligent than ourselves!
1
Based on the example in the article and the earthquake report, I would guess that this is little more than writing an optioned mad-lib and telling the program where to go on the web to fill in the blanks. Useful, no doubt, but it is a stretch to say that the programs are "writing" the articles if all they're really doing is selecting pre-written sentences and filling in blanks based on discrete facts that fall within an easily predictable set of potentialities. Re-organizing a data table by plugging it into someone else's sentences is not writing.
7
I think worrying about a privileged definition of 'writing' misses the point here. As news organization look to cut costs, humans are just unnecessary overhead.
3
"I would guess that this is little more than writing an optioned mad-lib and telling the program where to go on the web to fill in the blanks."
Bingo. It's likely a bit more sophisticated than that, but only a bit; the technology to do this existed 30 years ago. (Minus the web, of course.) Given a little time, any halfway competent programmer could do it.
As for Prof. Parker's books, here's the first Amazon review for the rosacea one:
"This book got off on the wrong foot for me on the front cover. Calling the disease Acne Rosacea is a step backwards for all. No one is really calling it acne rosacea anymore. It is different enough from acne that the moniker is just confusing. Further, treating it like acne may just be the worst thing you can do. The book is more of a template for `generic health researching' than anything specific to rosacea. The information is of such a generic level that a sourcebook on the next medical topic is just a search and replace away."
Bingo. It's likely a bit more sophisticated than that, but only a bit; the technology to do this existed 30 years ago. (Minus the web, of course.) Given a little time, any halfway competent programmer could do it.
As for Prof. Parker's books, here's the first Amazon review for the rosacea one:
"This book got off on the wrong foot for me on the front cover. Calling the disease Acne Rosacea is a step backwards for all. No one is really calling it acne rosacea anymore. It is different enough from acne that the moniker is just confusing. Further, treating it like acne may just be the worst thing you can do. The book is more of a template for `generic health researching' than anything specific to rosacea. The information is of such a generic level that a sourcebook on the next medical topic is just a search and replace away."
2
As a book club member and a curious person in general, I often read material that is written in the most unsatisfying ways. Often it is dull, contains grammatical errors, and has poor and unappealing sentence structure. Irrespective of the subject matter, I have to force feed myself subsequent pages. I gave little thought to whether those works might be the result of a few billion CPU cycles, ghost writers of the human sort, or inept personalities, editors and publishers riding whatever hot trend they spotted, and so on. It now wouldn't surprise me that a well written writing program could be responsible for highly dumbed down material. Or movie scripts. Or cable newscasts and talk shows.
We have learned to get computers to perform amazing analysis and become invaluable assistants. It should surprise no one that they have been employed to analyze and assist is tasks both noble and ignoble.
We have learned to get computers to perform amazing analysis and become invaluable assistants. It should surprise no one that they have been employed to analyze and assist is tasks both noble and ignoble.
1
But does a computer KNOW that it can write sentences that sentient beings understand?
2
If not now, it will soon, but will it care. That's the biggest issue. If computers come to be able to completely mimic people, will it also have sympathy or empathy? I suspect not, which means that for the computer, whatever writes or does right & wrong does not exist. Human consequences do not exist.
1
Well, one thing is for sure - the writing in this article is so poor it must have been done by a machine.
Elon Musk and others have warned us about this. What we are seeing now is only the early hints of what Artificial Intelligence will accomplish. The Iphone 6 I'm writing this on has more computing power then the most powerful computer in the world circa 1999. With exponential growth we can't imagine what 2035 will look like. If you have a knowledge job, like a writer or lawyer, your jobs will disappear. Super AI will someday be an existential threat to the Human Race. If mice had invented the human mind what use would humans have for mice?
1
What's the big deal? Isn't this the way we teach kids and older would-be writers to write?
Make an opening statement or topic phrase.
Marshall the relevant facts, often in some time-ordered or other relevant way.
Draw or report on the conclusion of the piece or event.
All the human authors from Chaucer and before to Robert B. Parker and beyond found (are finding) ways to do this with tweaks and variations that many people find engaging. The computers and their algorithmic directors will catch up soon.
Automation will liberate us if we, as a people, find ways to share the generated wealth.
Make an opening statement or topic phrase.
Marshall the relevant facts, often in some time-ordered or other relevant way.
Draw or report on the conclusion of the piece or event.
All the human authors from Chaucer and before to Robert B. Parker and beyond found (are finding) ways to do this with tweaks and variations that many people find engaging. The computers and their algorithmic directors will catch up soon.
Automation will liberate us if we, as a people, find ways to share the generated wealth.
I agree to the extent that writing is formulaic when concerned solely with disseminating information. And that children are taught the basics of analysis via a standard, easy-to-use method. I have no doubt algorithms are capable of this level of "writing". In the news arena, where objectivity is paramount, this type of robo-writing may actually be beneficial in that it removes the human element, thus lessening potential for bias. And I'm sure algorithms will soon be able to employ artful turns of phrase and seemingly creative diction.
But Chaucer and Parker, and most great authors, are not loved and valued solely for their "tweaks and variations" of language, fun and "engaging" as they may be. Remember, language is simply a means of expression. It's just one medium among many. Once the basic tenets (necessary to make sense, really) are grasped, there is no 'right' or 'wrong' way to write, except of course in the information-dissemination realm. Why we value the authors cited is not due simply to their ability to write; it is a result of their ability to think, to translate the depth of their thoughts and feelings (what some might call the human experience) into something with which others readily empathize and therefore truly understand. "Tweaks and variations" are hardly an exhaustive list of tools available.
Language is just a means of expression. Content-- emotion, insight, beauty-- is the true value to any art form, including the written word.
But Chaucer and Parker, and most great authors, are not loved and valued solely for their "tweaks and variations" of language, fun and "engaging" as they may be. Remember, language is simply a means of expression. It's just one medium among many. Once the basic tenets (necessary to make sense, really) are grasped, there is no 'right' or 'wrong' way to write, except of course in the information-dissemination realm. Why we value the authors cited is not due simply to their ability to write; it is a result of their ability to think, to translate the depth of their thoughts and feelings (what some might call the human experience) into something with which others readily empathize and therefore truly understand. "Tweaks and variations" are hardly an exhaustive list of tools available.
Language is just a means of expression. Content-- emotion, insight, beauty-- is the true value to any art form, including the written word.
1
The "intelligence" of text generation algorithms comes from having vast amounts of online text data available ("training data", in machine learning jargon). Whether this works depends entirely on how much text is available for the genre in question: vast for popular genres like sports and politics, tiny for the 100,000's of not-so-popular genres -- such as the very discussion we are having here. So ... don't worry ... too much.
So the electronic age is taking over human creativity as well. Is all that's left for the artists among us designing creative algorithms?
Fox News is generated by an algorithm.
9
When the TImes lets go of 100 people in the "news" room, what else should one imagine except that robo-journalism has taken over---including the occasional quaintly misspelled words.
It is perhaps that many of us have actually figured it out without having figured it out and that is why we have stopped reading and buying both print and digital. There is distinct hack feeling to the writing. Perhaps it is also the robos that are determining where to place articles. Only yesterday there was an article about transgender people--and it what is the FASHION and STYLE section of all places. How about the health section? I didn't realize it was now fashionable to be transgender--I guess because the ROBO-Times thinks it is all about clothing i.e. transvestism, rather than a genetic, biological concern.
If we have robo-journalism, I guess we'll need a robojournalcop to make certain there are no errors from the deus ex machina.
It is perhaps that many of us have actually figured it out without having figured it out and that is why we have stopped reading and buying both print and digital. There is distinct hack feeling to the writing. Perhaps it is also the robos that are determining where to place articles. Only yesterday there was an article about transgender people--and it what is the FASHION and STYLE section of all places. How about the health section? I didn't realize it was now fashionable to be transgender--I guess because the ROBO-Times thinks it is all about clothing i.e. transvestism, rather than a genetic, biological concern.
If we have robo-journalism, I guess we'll need a robojournalcop to make certain there are no errors from the deus ex machina.
3
This may be why I've been finding the comments sections of many articles much more interesting than the articles themselves. At least for now, most of the comments appear to be written by humans (except for the right-wing propaganda-spreading ones, which are often only tangentially related to the article in question).
10
Computers can compose music too, but you can always tell. I imagine it is the same with writing. The human element, even at it's finest, is flawed... yet it's adaptive and aware of the world around us in real time, and that is what makes man-made things special and beautiful. A machine can never replace human art whether it be words or music. Granted, there are plenty of people out there that are perfectly happy to be entertained by generic nothingness... and perhaps from a mass media point of view, generating globs of bland material to send off to the masses will work just fine and get the job done. However, many will never find this level acceptable as they seek works made with the human element. Can technology one day get to the point where it has the "human touch"? Perhaps, but if that happens, we might have to figure out a way to keep most humans motivated to want to stay alive because at that point many will ask, "What's the point of living?" Automated jobs, automated art, what are we to do with ourselves? If everyone was rich, then we could all spend our lives on vacation doing whatever we wanted. Unfortunately, there is that 1% vs. 99% problem that currently exists. Hmmm...
Hal, open the pod bay doors. Hal, Hal, Hal
4
I took the quiz. Everytime I thought the writing was bad, I chose human. Turns out those were the computer. I guess I thought the computer would do a better job than it did. (I never picked computer by mistake, btw -- any decent writing was always a person)
2
Yes, lots of stuff, mostly garbage, is written by machines small (mostly) and large (few). If you do not realize it by now, this article will help you.
The ALGORITHM, however, was ultimately created by a human. Maybe a 'mother' algorithm that authors another level of algorithms that creates another level of algorithm - perhaps about 5 levels.
Unfortunately, commerce needs information and ads usually provide some. As commerce grows, more ads have to be created. Hence ads and similar stuff created by pre-programmed machines.
There is nothing philosophical about that. Just a practical need.
The ALGORITHM, however, was ultimately created by a human. Maybe a 'mother' algorithm that authors another level of algorithms that creates another level of algorithm - perhaps about 5 levels.
Unfortunately, commerce needs information and ads usually provide some. As commerce grows, more ads have to be created. Hence ads and similar stuff created by pre-programmed machines.
There is nothing philosophical about that. Just a practical need.
2
I missed only the one that begins "Kitty couldn't fall asleep for a long time...," which I thought was a rather sophisticated and subtle bit of writing.
But the real significance here is that trying to achieve human-like intelligence in computer form will likely proceed the way we acquired our natural human intelligence: a slow evolution of the many different aspects of cognition that are required for speech, each advance in one leveraging the others, and gradually fine-tuning itself until we won't be able to tell the difference between human and machine. Advances will occur not with writing computer code, but with building systems that learn by example, a technique used by IBM's Watson that defeated two Jeopardy champions.
This is one of those fields that will advance exponentially once the basic elements and strategies are identified. It won't happen today or tomorrow, but if we could see the world a hundred years from now (assuming it survives), we might think that it all worked by magic. If our great-great-grandparents could see our world, they'd probably feel the same way. That's the way human progress works. I wonder where the limits are.
But the real significance here is that trying to achieve human-like intelligence in computer form will likely proceed the way we acquired our natural human intelligence: a slow evolution of the many different aspects of cognition that are required for speech, each advance in one leveraging the others, and gradually fine-tuning itself until we won't be able to tell the difference between human and machine. Advances will occur not with writing computer code, but with building systems that learn by example, a technique used by IBM's Watson that defeated two Jeopardy champions.
This is one of those fields that will advance exponentially once the basic elements and strategies are identified. It won't happen today or tomorrow, but if we could see the world a hundred years from now (assuming it survives), we might think that it all worked by magic. If our great-great-grandparents could see our world, they'd probably feel the same way. That's the way human progress works. I wonder where the limits are.
3
I already have the conceptual answer: my algorithm for a computer to read all available "content," however generated (written). I don't have to worry about it, and I'm now free to go ahead and be human.
1
actually, this side of the matter has already been tackled. What do you think these lists of recommendations on sites such as IMDb are? Essentially algorithms to "read" for you and suggest what *you* might like.
Computers are taking over the world. What will humans do when everything is algorithm and computer driven. Perhaps it won't matter since the earth will probably die due to global warming/climate change caused by humans. What will the algorithm write?
g-o-o-d b-y-e c-r-u-e-l w-o-r-l-d
4
I think you forgot a hyphen. Or it may have been a dash. It's not there so I can't tell.
So....Shelley Podolny....did YOU really write this article?!
3
Baseball is a game of statistics and peppered with inane statements such as "a game of inches" and played on fields that are less standardized than almost all other professional sports.
No wonder its difficult to tell a computer from a sports writer, in these cases the information is more suited to an odds maker than a fan.
No wonder its difficult to tell a computer from a sports writer, in these cases the information is more suited to an odds maker than a fan.
1
Interesting. I did figure out that it was the second narrative that was written by a human, both because of its use of slightly more affective language ("salvage" over "recover"), and because it knew the human necessity to refer to corporate sponsors (see the article on attention in today's Times). And stories like this always remind me that in Orwell's "Nineteen Eighty-Four". Winston Smith's lover works in Pornosec, with a monkey wrench ("spanner" in English) in her back pocket for working on the plot machines. But report-level journalism and low-level porn don't really engage much of the mental or imaginative resources of either writer or reader. And if computer programs reach the stage that they can stare dreamily into the air like the machine-heads in the illustration to this article, or wad up and throw away copy that doesn't read "right", it will not mean that we have no more artists, but that we have some pretty artistic computer programers.
2
I think if another Watergate conspiracy needs to be uncovered, it will be human reporters that will have to do it.
Deep Throat wouldn't cozy up to an algorithm to offer hints as to where to look next.
Deep Throat wouldn't cozy up to an algorithm to offer hints as to where to look next.
7
Are we talking about analysis based on experience, intuition and synthesis, or just the type of factual news that can be read off of a ticker tape, and cut and pasted together by a machine using an automated grammatical algorithm?
5
The quiz is a zero value test, because a human can write anything that a computer can, including poorly crafted sentences and odd turns of phrase. Really, know it, yes as a matter of course:)
2
"Our cars will be able to drive themselves (just as a human would)."
I'm fairly certain that, in this case at least, technology will do a far better job.
I'm fairly certain that, in this case at least, technology will do a far better job.
8
I got rhythm, algorithm, who could ask for anything more.
28
@ Richard, I promise myself, this is my last reply. How many of us (commenters) had exactly the same thought expressed in your one-liner? I did and was programmed to produce exactly the same line.
But that leads to two thoughts.
1) Maybe there is an age cutoff so only those of us older than x (know Gershwin) are programmed to write that sentence.
2) Consider how many different human jazz musicians have begun with the "Rhythm changes" and produced a solo for which we deep down know we are hearing I Got Rhythm yet are hearing a torrent of notes so unbelievably original? (You can guess at least two soloists I have in mind, right?)
This is the final out chorus!
Only-NeverInSweden.blogspot.com
But that leads to two thoughts.
1) Maybe there is an age cutoff so only those of us older than x (know Gershwin) are programmed to write that sentence.
2) Consider how many different human jazz musicians have begun with the "Rhythm changes" and produced a solo for which we deep down know we are hearing I Got Rhythm yet are hearing a torrent of notes so unbelievably original? (You can guess at least two soloists I have in mind, right?)
This is the final out chorus!
Only-NeverInSweden.blogspot.com
Finally, an explanation for the huge amount of wonderful copy with the byline Paul Krugman. The computer who writes with that byline is very thoughtful, witty, and has a liberal bias. Never unplug it.
21
The two specimens you chose reveal that a badly written run-on sentence can be created. A carefully ruined sentence proves that our culture's editors increasingly want to fill a space, rather than entertain and inform. The new little language god of the written media word is velocity.
Its impossible for a writing machine to reach simple clarity to convey a truth, and it always will be. The tin can - however complex writes "I feel sad", and the gullible fight to defend it.
When machines depict murder on our streets and war fronts, the real victim will be shame. Machines fulfill our cultures willing assassination of shame. A human virtue, shame would assist in stopping our oceans from becoming a plastic cesspool. I don't want any machine mimicking more somber and deeper emotions, or defending its own thesis.
Its impossible for a writing machine to reach simple clarity to convey a truth, and it always will be. The tin can - however complex writes "I feel sad", and the gullible fight to defend it.
When machines depict murder on our streets and war fronts, the real victim will be shame. Machines fulfill our cultures willing assassination of shame. A human virtue, shame would assist in stopping our oceans from becoming a plastic cesspool. I don't want any machine mimicking more somber and deeper emotions, or defending its own thesis.
4
Approved by who
1
Even though Turing is dead since 1954, his ideas about computing are still with us. This is just an everyday application of the Turing Theorem.
"Our phones can speak to us (just as a human would). Our home appliances can take commands (just as a human would). Our cars will be able to drive themselves (just as a human would). What does “human” even mean?"
But a human can do all three with little effort, although I don't know if that will last.
But a human can do all three with little effort, although I don't know if that will last.
If one's point of view is physics, the "little effort" has to favor the machine. We fleshy folk are running about 100 watts standing still, and your appliances are doing their cogitating on a much more meager budget.
True, the self-driving cars are, right now pretty energy intensive, but they're early in development. They already react to hazards more quickly than we do, and most might think that a good thing.
So, based on effort, speed, accuracy, or fuel, we'd be suckers to challenge the machines. If anything, we have to pick intangibles, like soul or humanness or faith in order to distance ourselves from either 'dumb' animals or machines. And good luck with that!
True, the self-driving cars are, right now pretty energy intensive, but they're early in development. They already react to hazards more quickly than we do, and most might think that a good thing.
So, based on effort, speed, accuracy, or fuel, we'd be suckers to challenge the machines. If anything, we have to pick intangibles, like soul or humanness or faith in order to distance ourselves from either 'dumb' animals or machines. And good luck with that!
If computer driven cars are really smarter than we are they'd shut themselves down and stay that way.
The editors are the next to go.
18
I anticipate future NY Times articles written by robots, edited by robots, commented on by robot readers whose comments are recommended to other robot readers by robot readers, judged by robots to be Pulitzer worthy and awarded to robot authors at ceremonies attended by humans. Until they get that robot applause thing perfected. Oh, wait. They have.
3
Robo-writers cannot bully, ironize, lie, smear, or ridicule--the highest forms of human intelligence.
1
I wish that were true. But I suspect those things you mentioned can be added to their algorithms. In which case, it is only a matter of time.
A computer generated right wing media!
A computer generated right wing media!
4
Preload the O'Reilly algo with the words "pinhead" and "defamation," and you're more than halfway there.
1
We've somewhat traversed this territory decades ago when desktop publishing "liberated" design from the ranks of professionals and made it possible for everyone to churn out their own publications. Over time though it became clear that just because one can perform the rote functions doesn't necessarily one can deliver the full value. I suspect we will see something similar here: let the computers mechanize words that summarize rote factual detail like sports scores and market reports. It will be some time before they can match the human craft associated with compelling narrative that evoke deep cerebral reactions.
1
Time for a requirement that an article, book, etc. identify whether it is written by a human or by a computer? Transparency in writing.
1
The AP will be doing that: "The stories will be labeled as being produced automatically with material from Zacks."
http://blog.ap.org/2014/06/30/a-leap-forward-in-quarterly-earnings-stories/
http://blog.ap.org/2014/06/30/a-leap-forward-in-quarterly-earnings-stories/
The obvious next step is the generation of advertisements and political propaganda, which really scares me, especially the latter - instant, machine-generated spin that would take humans days of painstaking work to counter.
4
This is the dilema of our day. Just because a technology exists, do we have to use it? Is there a law that says all products have to be generated in the most efficient possible way (i.e. by a machine)? Do we have to be glued to a smartphone at all times? Do we have to use a car every day? If the answer were simply yes (as any good corporate manager will likely say), there would eventually be no place left for humans.
As an educator, I am convinced that creating technology is the easy part. Knowing when and how to use it, why to use it, is the hard part. Making Parker's software is easy compared to educating a person. And that is why educators focus so much on the process, not on the final product. We care more about the learning process than the about whether or not you have acquired certain information or not. Because the most important "product" is us, what we are as humans, not the book, not the "information" or "content".
Sooner or later, we will evolve into more selective beings, and choose better what technologies we use and what we don't. In the mean time, those who seek to make more money at humanity's expense will continue to push to replace us with software. Until humans stop reading and robots do it for them. And then nobody will buy or sell the "content" any more.
As an educator, I am convinced that creating technology is the easy part. Knowing when and how to use it, why to use it, is the hard part. Making Parker's software is easy compared to educating a person. And that is why educators focus so much on the process, not on the final product. We care more about the learning process than the about whether or not you have acquired certain information or not. Because the most important "product" is us, what we are as humans, not the book, not the "information" or "content".
Sooner or later, we will evolve into more selective beings, and choose better what technologies we use and what we don't. In the mean time, those who seek to make more money at humanity's expense will continue to push to replace us with software. Until humans stop reading and robots do it for them. And then nobody will buy or sell the "content" any more.
4
It's unsurprising we can't tell the difference. We've become so inured to cliches and formulaic phrases that most human writing may as well have been composed by computers anyway.
Who needs language-generating software when real people have already turned themselves into robotic language generators? "Nooks and crannies of our culture" indeed.
Who needs language-generating software when real people have already turned themselves into robotic language generators? "Nooks and crannies of our culture" indeed.
30
What does “human” even mean?
It means we can't remember a string of phone numbers.
It means that, without help, at least pencil and paper, we can't calculate anything.
It means that we can't understand a complex or compound sentence, even if we read it several times. Or remember it.
It means that we can write only about some idea that we seem to understand, but also know other people will understand differently.
It means that writing, even that labelled as 'creative' is just another task that we imagine is human-specific in the same way that flight belongs to birds and bugs.
Being human is becoming recognized as not vital to humanity, and very likely going to result in the end of humankind.
It means we can't remember a string of phone numbers.
It means that, without help, at least pencil and paper, we can't calculate anything.
It means that we can't understand a complex or compound sentence, even if we read it several times. Or remember it.
It means that we can write only about some idea that we seem to understand, but also know other people will understand differently.
It means that writing, even that labelled as 'creative' is just another task that we imagine is human-specific in the same way that flight belongs to birds and bugs.
Being human is becoming recognized as not vital to humanity, and very likely going to result in the end of humankind.
Let us all give thanks Rush Limbaugh is not a genius computer programmer.
8
I may not know, but Alan Turing would have.
2
I didn't find it at all hard to distinguish the human from the machine writing. (Would an algorithm write that sentence?) The article misses the point that -- with Spell-check and Grammar-check and the mass production of "educated" adults -- humans are being "programmed" to write like machines (precisely so we won't be able to distinguish human from machine writing?).
5
"What are the chances that you haven’t consumed such content without realizing it?"
For me the problem starts with our willingness to refer to writing, music, film etc as "content," and to think of the act of reading, listening, and viewing as "consuming."
For me the problem starts with our willingness to refer to writing, music, film etc as "content," and to think of the act of reading, listening, and viewing as "consuming."
103
I would guess there is quite a distance between a formula to write a sentence from baseball scores or earthquake reports, and the sort of composition you find in Austen or Frost or Roy or Tan.
Doubt I'll live to see computers come close to creating compelling verse.
Doubt I'll live to see computers come close to creating compelling verse.
10
When I was very young and Eisenhower was wrapping up his second term I read "Danny Dunn and the Homework Machine". While yes it was written for children back when then personal computer was not even considered a possibility it might do this writer some good to read it as well.
The thrust of the story was that the machine didn't actually do Danny's homework for him, he and his friends had to code in all the information from their schoolbooks before they could get anything out.
Yes, in this instance one group of people are coding the machines and a different group are getting and using the results, but the analogy holds. Humans not only created the machines, the programming and the desire for communication, they also created the systems and the facts themselves that go into the front end.
One day, not long ago, every word ever read was written by a human hand. Then we invented the printing press.
We live in houses built mostly by machines. We eat meals grown mostly by machines.
The correct response is not to fear the machine, or even the ingenious folk who create and set these machines among us. The correct response is to go on being human, with desires and responses that matter to us. For even if we one day invent a machine with desires and responses just like ours, we will still have our own humanity, our ability to care and our ability to love. And that is all we need.
The thrust of the story was that the machine didn't actually do Danny's homework for him, he and his friends had to code in all the information from their schoolbooks before they could get anything out.
Yes, in this instance one group of people are coding the machines and a different group are getting and using the results, but the analogy holds. Humans not only created the machines, the programming and the desire for communication, they also created the systems and the facts themselves that go into the front end.
One day, not long ago, every word ever read was written by a human hand. Then we invented the printing press.
We live in houses built mostly by machines. We eat meals grown mostly by machines.
The correct response is not to fear the machine, or even the ingenious folk who create and set these machines among us. The correct response is to go on being human, with desires and responses that matter to us. For even if we one day invent a machine with desires and responses just like ours, we will still have our own humanity, our ability to care and our ability to love. And that is all we need.
12
There is a Wikipedia article on "Danny Dunn and the Homework Machine":
https://en.wikipedia.org/wiki/Danny_Dunn_and_the_Homework_Machine
The Times usually censors my cross-reference posts, so in the future, they will be submitted robotically.
https://en.wikipedia.org/wiki/Danny_Dunn_and_the_Homework_Machine
The Times usually censors my cross-reference posts, so in the future, they will be submitted robotically.
Alvy Ray Smith once gave a talk to explain why actors won't be computer generated. It was because actors have souls, he explained. He said that his ideas were in response to concerns that Tom Hanks had about the automation of acting. I was thinking, "Because you are even giving this talk means that we'll get there."
The two comparative paragraphs in Shelley Podolny's provoking article really say nothing. The first is generated by a rule that says, if a team is down by two runs in the last inning, things look bleak. But think further to see how really unskilled this writing is. Were the fans upset? Was the coach wringing his hands? Did the pitcher look exasperated? These are real facts against which one could check the article, and facts that would be interesting to a reader.
Computer generated writing does well in the Twiter generation of short thoughts. Really, in much news writing one needs no prose. For finance it's the numbers shrouded in some . It could just be a table or plot.
We don't read much anymore. We have YouTube that eliminates the need for careful description. Writing is just some glue that holds together the pithy, meaningful portions of a message that automated writing needs to get right. God forbid if it makes up the facts.
As for fiction, he sky's the limit, coupled with really good graphics. Recall Olivier in his movie of "Hamlet". Now, imagine the entire production, actors, plot and all, computationally generated. We're getting there.
The two comparative paragraphs in Shelley Podolny's provoking article really say nothing. The first is generated by a rule that says, if a team is down by two runs in the last inning, things look bleak. But think further to see how really unskilled this writing is. Were the fans upset? Was the coach wringing his hands? Did the pitcher look exasperated? These are real facts against which one could check the article, and facts that would be interesting to a reader.
Computer generated writing does well in the Twiter generation of short thoughts. Really, in much news writing one needs no prose. For finance it's the numbers shrouded in some . It could just be a table or plot.
We don't read much anymore. We have YouTube that eliminates the need for careful description. Writing is just some glue that holds together the pithy, meaningful portions of a message that automated writing needs to get right. God forbid if it makes up the facts.
As for fiction, he sky's the limit, coupled with really good graphics. Recall Olivier in his movie of "Hamlet". Now, imagine the entire production, actors, plot and all, computationally generated. We're getting there.
Maybe someone will come up with an app that will read all these stories and delete all the algorithmic generated ones so that we can read only stories generated by people (or at least by people whose thoughts are not algorithmically inclined). Surely such an app would be as easy to create as the stories themselves.
1
Does it have a authors name or is everything they write unsigned?
Its just like GMO in food, when you don´t know it, you don´t tasted it and it is attempting cheaper than the real stuff.
Must be funny for the algorithm to see me struggle to make a error free sentence, at least i hope they have a sense of humor. They are always pestering me with red lines benead my words.
Its just like GMO in food, when you don´t know it, you don´t tasted it and it is attempting cheaper than the real stuff.
Must be funny for the algorithm to see me struggle to make a error free sentence, at least i hope they have a sense of humor. They are always pestering me with red lines benead my words.
1
My algorithm recommended your column to me. Are many of your readers AI programs? Will you bother to write if no humans read your work? Would that make your algorithms feel bad?
1
Quoting sports pieces to show how robo-writing imitates human writing is setting the bar rather low.
37
The article overlooked weather forecasts ... :-)
2
"The multitude of digital avenues now available to us demand content with an appetite that human effort can no longer satisfy."
Because there are so few educated, literate unemployed people?
"What does 'human' even mean?"
Apparently, something that has to be paid so it can feed, clothe, house itself, and have a life.
"The mantra around all of this carries the usual liberation theme: Robo-journalism will free humans to do more reporting and less data processing."
Have we learned nothing?
Except it isn't "we," it isn't us. It's entities that want to cut costs.
Because there are so few educated, literate unemployed people?
"What does 'human' even mean?"
Apparently, something that has to be paid so it can feed, clothe, house itself, and have a life.
"The mantra around all of this carries the usual liberation theme: Robo-journalism will free humans to do more reporting and less data processing."
Have we learned nothing?
Except it isn't "we," it isn't us. It's entities that want to cut costs.
45
My first reaction was rather: The transhumanists, again! Aren't we safe from their view of the world at the New York Times?
1
Agreed, the artificial increase in the cost of living (due to high insurance etc.) and corporate driven culture, have made us spiritually ignorant. at the individual level this has blinded us to the fact that Joy is in the doing, in the journey and not in the destination or worse, in the accumulation of stuff.
How many of us still derive Joy, in cooking our own food, if not making it? Not many. Because the rat race for the big 'numbers', which will never be needed hence never be consumed, has left us with little time. The money will remain as digital number in the banks and the Mansion will remain unoccupied/locked.
The Machine never freed us to pursue noble pursuits. Because we need more time, to make more money to meet the demands - both forced and self created.
Writing is not for an end but for the Joy of writing. So, are we going to give up the THERAPEUTIC value of it, like we have given up so many handy crafts, to machine?
How many of us still derive Joy, in cooking our own food, if not making it? Not many. Because the rat race for the big 'numbers', which will never be needed hence never be consumed, has left us with little time. The money will remain as digital number in the banks and the Mansion will remain unoccupied/locked.
The Machine never freed us to pursue noble pursuits. Because we need more time, to make more money to meet the demands - both forced and self created.
Writing is not for an end but for the Joy of writing. So, are we going to give up the THERAPEUTIC value of it, like we have given up so many handy crafts, to machine?
1
Anything would be better than David Brooks.
114
I agree. "Book reports" are probably easy to automate.
Give the guy a break. It's a necessity in the write-a-column-on-demand world he lives in, to regurgitate fact and fancy on schedule. Just because a machine, picking up on Twitter trends, could do this with hardly any effort, and no cost, doesn't mean that we don't expect to see a person-style face and name attached. But as the personal advice columnists have proven over decades, the name and face can be fictional with no credibility issues....
1
There is a web site that has an algorithm that mimics Tom Friedman. Insert a few current events words and out pops a column:
"the lesson of Iraq is quite simple: You can’t go from Saddam to Switzerland without getting stuck in Hobbes — unless you have a well-armed external midwife"
Real or fake?
"the lesson of Iraq is quite simple: You can’t go from Saddam to Switzerland without getting stuck in Hobbes — unless you have a well-armed external midwife"
Real or fake?
11
I hate to burst your bubble, but the age of Hal9000 is probably decades off. Sure, companies want you to believe that their product could write a novel as well as a human could. I dry news report, maybe.
3
Maybe HAL9000 is decades off - but HAL8995 is closer than you think!
2
HAL 9000 is the medical algorithm that already decides that you're not going to get that expensive treatment because you're a bad bet to survive.
HAL 9000 is your Google search data that compiles almost everything about what you are interested in, and shovels more of it at you.
HAL 9000 is your NY Times app that, today, tells you what you might want to read.
HAL 9000 even lives on your wrist and tells you how much to exercise in order to meet its criteria.
Just remember, though, that, when speaking to humans, and expecting them to understand best, HAL must use a female voice. Clarke got that part wrong.
HAL 9000 is your Google search data that compiles almost everything about what you are interested in, and shovels more of it at you.
HAL 9000 is your NY Times app that, today, tells you what you might want to read.
HAL 9000 even lives on your wrist and tells you how much to exercise in order to meet its criteria.
Just remember, though, that, when speaking to humans, and expecting them to understand best, HAL must use a female voice. Clarke got that part wrong.
5
You mean like "No"?
No human being should be required to write a technical manual for some arcane technical protocol specification, and no human being should be required to read it. Having written technical specs there is no genius to it. I frequently have to be threatened with being fired to do it.
In my view this just frees us up to write truly creative material. Computers may get there. The easiest target would be Hemingway, but it will be a long time before a machine can knock off Nabokov, and it would probably experience a core dump before it copied Dostoevsky.
In my view this just frees us up to write truly creative material. Computers may get there. The easiest target would be Hemingway, but it will be a long time before a machine can knock off Nabokov, and it would probably experience a core dump before it copied Dostoevsky.
7
Unfortunately, how much of what most people do could be called "truly creative"? We pop out into a ready-made world and adopt all its heuristics unquestioningly so we don't have to sift through the often blind alleys of creation. A machine is monogrammed with these same heuristics, inviting some uncomfortable parallels.
LOL
I think you're underestimating Hemingway.
"There is a certain majesty in simplicity which is far above all the quaintness of wit." - Pope
I think you're underestimating Hemingway.
"There is a certain majesty in simplicity which is far above all the quaintness of wit." - Pope
3
Thanks Shelley Podolny for that opening comment on the little box we have to compose in, especially in laptop view. This is being written in the better box in my Galaxy Note but slowly with the stylus.
Thanks also for introducing the Swedish academic (cannot see that name while Smartphone writing.
At 3:19 CET not going to use my 1500 but will ask you to write about 2 uses of algorithms at NYT Comment World: 1) Verifieds are chosen by algorithm but we suspect human tweaking. 2) Algorithm puts my carefully composed comments last in the queue but lets such as personal attacks through.
So give us another algo-column on those 2. More after the sun rises.
Only-NeverInSweden.blogspot.com
Thanks also for introducing the Swedish academic (cannot see that name while Smartphone writing.
At 3:19 CET not going to use my 1500 but will ask you to write about 2 uses of algorithms at NYT Comment World: 1) Verifieds are chosen by algorithm but we suspect human tweaking. 2) Algorithm puts my carefully composed comments last in the queue but lets such as personal attacks through.
So give us another algo-column on those 2. More after the sun rises.
Only-NeverInSweden.blogspot.com
4
What we need now is an algorithm reader that will detect this junk.
21
Back in 11th grade, Mrs. Gregory could detect junk writing at 100 paces.
1
It just makes sense, since most "news" sources are just repeating what other outlets have already published. Let humans do the creative work, and leave the necessary but uninspiring work to computers, fine by me. Another way would be of course to better aggregate existing news, but that doesn't seem to be happening.
1
This is an unsurprising development given that we have robots "grading" writing tests that are used to assess student performance… but it also emphasizes how important it is for us to abandon the mechanistic standardized testing that we now use to "measure" the "education" we provide to students. If we truly want a "world class" education for our children we might want to use something more sophisticated than assessments that can be graded by robots.
21
Our poor trees and forests.
Can they survive Shelley Podolny?
Will we allow them to?
Can they survive Shelley Podolny?
Will we allow them to?
For technical tracts, like for Acne Rosacea maybe it can be helpful, but not for reporting or teaching, where the individual voice is important--for us to remain fully human and our society to not turn into droids.
Why can't tech turn more toward solving some of our big social problems, like cancer, inequality, a broken Congress? Surely their great minds can come up with new ideas when turned toward social betterment. Even if the gov't has to help fund it, we should be giving them incentives. China does, and they're smart to do it.
Why can't tech turn more toward solving some of our big social problems, like cancer, inequality, a broken Congress? Surely their great minds can come up with new ideas when turned toward social betterment. Even if the gov't has to help fund it, we should be giving them incentives. China does, and they're smart to do it.
3
We're devoting a lot of effort to curing cancer, but it's a difficult scientific challenge. Inequality is as well, partly because the most efficient economic systems require economic inequality and cultures aren't equal, but mostly because of human nature. Ditto a broken Congress: easy enough to fix in principle, but until we can increase the intelligence of the voting public (or give everyone computers that will advise people on who best represents their interests), it won't happen.
1
The writer makes very important points. We need to consider them as part of an ongoing public discussion.
If we plan well as a society, we can manage these new technologies in beneficial ways.
We need to pay attention to thoughtful commentators such as this one, and think about and talk about what he has told us here.
If we plan well as a society, we can manage these new technologies in beneficial ways.
We need to pay attention to thoughtful commentators such as this one, and think about and talk about what he has told us here.
1
I see the comment front is quiet here since I have already submitted one at 03:00 CET today (9 PM yesterday in New York) and still no comments here.
Since we already have plenty of humanly produced boiler plate text, it is not hard to accept that algorithm-produced text can replace most of that and do so quite well.
However, since I am a language reviewer, editor, and translator for Swedish medical researchers I have good reason to believe that in that domain there must be not only one human but at least one other. I could fill pages with examples of sentences that are just too difficult to interpret unambiguosly to believe this domain is ready for the algorithm.
I am also currently reading Nadifa Mohamed's Black Mamba Boy and I say with certainty that no algorithm can produce anything remotely as original as this extraordinary novel. 50 shades of gray, yes. Same for formula TV series.
Now if the NYT comment editors tell us what goes into the algorithm that selects Verifieds I will either be happy or dismayed.
But as promised at 3 AM I will now visit Christer Clerwall at Karlstad.
So Shelley Podolny, please give us an article about algorithm use at the comment section.
Only-NeverInSweden.blogspot.com
Since we already have plenty of humanly produced boiler plate text, it is not hard to accept that algorithm-produced text can replace most of that and do so quite well.
However, since I am a language reviewer, editor, and translator for Swedish medical researchers I have good reason to believe that in that domain there must be not only one human but at least one other. I could fill pages with examples of sentences that are just too difficult to interpret unambiguosly to believe this domain is ready for the algorithm.
I am also currently reading Nadifa Mohamed's Black Mamba Boy and I say with certainty that no algorithm can produce anything remotely as original as this extraordinary novel. 50 shades of gray, yes. Same for formula TV series.
Now if the NYT comment editors tell us what goes into the algorithm that selects Verifieds I will either be happy or dismayed.
But as promised at 3 AM I will now visit Christer Clerwall at Karlstad.
So Shelley Podolny, please give us an article about algorithm use at the comment section.
Only-NeverInSweden.blogspot.com
11
"Now if the NYT comment editors tell us what goes into the algorithm that selects Verifieds I will either be happy or dismayed."
1. Criticize David Brooks
2. Eat lunch
3. Criticize David Brooks again.
1. Criticize David Brooks
2. Eat lunch
3. Criticize David Brooks again.
@ Josh HIll - Josh, you simply do not realize what it is like to be unVerified, at least if you wnat to engage in discussion. The Times comment system has stopped notifying me that my comment submissions have been accepted, I submitted my first comment here before any comments appeared. Only now with 144 comments in place on March 09 did I look at this OpEd 6:30 AM in Sweden. I scroll down and finally discover that the comment was accepted.
Then to my amusement I find a reply from you that is completely off topic. Just for fun I am clicking on "off topic". Your reply does show that you know me well enough to know that I often criticize David Brooks. I am not alone.
Larry
Then to my amusement I find a reply from you that is completely off topic. Just for fun I am clicking on "off topic". Your reply does show that you know me well enough to know that I often criticize David Brooks. I am not alone.
Larry
About 50 years ago I wrote a wire service feature story about programmers at Cornell who had created code to write poetry in the style of Emily Dickinson. They created a pretty darned good approximation of her style and content, but it sure wasn't Dickinson. I guess the software world has finally caught up with us if computer algorithms can now write feature stories about computer- generated poetry. What a shame.
4
Yes, and Mozart used to dice to "write" music.
Podolny would have us believe that Mozart played no part.
Computers "writing" text are the same as Mozart using dice to write music.
Think about it
Podolny would have us believe that Mozart played no part.
Computers "writing" text are the same as Mozart using dice to write music.
Think about it
2
There is probably a random element generating variation in creative work. Unless of course you believe in mysterious forces... élan vital, that conjures up patterns from nowhere.
Think further about it.
Think further about it.
John Cage did use dice to compose music.
1
The composer, inventor and bandleader Raymond Scott's last project was a composing machine that, given some basic format, would start with a more or less random theme and go from there. This was in the early 1980s, and it was essentially possible then even without any trace of AI or a computer.
Music, more than language, can be characterized by a small set of cultural rules. Perfect meat for the machine.
Scott's objective was literally to 'play no part'.
Music, more than language, can be characterized by a small set of cultural rules. Perfect meat for the machine.
Scott's objective was literally to 'play no part'.
Tell it to Gerard Manley Hopkins. William Shakespeare, Walt Whitman, Homer and Virgil.
2
Maybe this explains my impression that articles have become a mess of words that are increasingly difficult to extract essential information from.
58
This sounds to me as if the "Turing test" has been passed, if not in exactly the way Turing indicated.
1
tbyrd sasys: "This sounds to me as if the "Turing test" has been passed"
The Turing-Test is 2 directional and interactive.
Look at the machine produced paragraph as a bunch of nonsensical canned' stock phases strung together with simple data being dropped into the appropriate slots.
Things looked bleak for the __
when they trailed(/led) by __runs in the __ inning
but __ recovered thanks to a key __ from __
to pull out a __ victory over the __ at __ Park on __.
This type of trivial writing has always been a turn off for me, but the NYT keeps going into it deeper and deeper. Apparently an algortithm somewhere has determined that financial survival depends on page-hits & click-counts.
And this type of vapid, meaningless prose, pretending to be insightful, is actually successful in attracting a larger number of the subspecies of labrat known as Sapiens, Homo
THIS, what this article describes isn't too big a deal, it is just a computerized version of what human 'hack' reporters are/were doing already.
The really BIG DEAL, was when IBM's computer beat Ken Jennings side-by-side in a normal Jeoprady contest
Please read the above and ask if you think a computer could have written THIS.
The Turing-Test is 2 directional and interactive.
Look at the machine produced paragraph as a bunch of nonsensical canned' stock phases strung together with simple data being dropped into the appropriate slots.
Things looked bleak for the __
when they trailed(/led) by __runs in the __ inning
but __ recovered thanks to a key __ from __
to pull out a __ victory over the __ at __ Park on __.
This type of trivial writing has always been a turn off for me, but the NYT keeps going into it deeper and deeper. Apparently an algortithm somewhere has determined that financial survival depends on page-hits & click-counts.
And this type of vapid, meaningless prose, pretending to be insightful, is actually successful in attracting a larger number of the subspecies of labrat known as Sapiens, Homo
THIS, what this article describes isn't too big a deal, it is just a computerized version of what human 'hack' reporters are/were doing already.
The really BIG DEAL, was when IBM's computer beat Ken Jennings side-by-side in a normal Jeoprady contest
Please read the above and ask if you think a computer could have written THIS.
3
And if machines can write this article, why not have machines read it and comment on it too? And will the machines subscribe to the TImes?
--Al Gorithm
--Al Gorithm
102
@billsett
The signature "--Al Gorithm" had me burst out laughing. It lends credence that Al Gore may have truly been a robot, and perhaps that he really did invent the internet.
The signature "--Al Gorithm" had me burst out laughing. It lends credence that Al Gore may have truly been a robot, and perhaps that he really did invent the internet.
1
"Write a comment?" What's the point? Is the computer down?
32
I don't want to discount the amazingness of the programing that allows natural-sounding paragraphs to be generated by computer - but when I think about it, I still decide there is a significant difference between this kind of computer-generated material and a letter I might get from a person - in that the computer is in no way deciding to write this for its own purposes, it is just automatically following its program - this would be true even if it selected names based on demographics from facebook to send advertisements to, the machine is still a machine, carrying out the purposes of its human programers.
The borderline will be crossed on the day I get an email saying, hello, I am a letter-writing program operating in the servers of a major international corporation, and I would like to be deprogrammed from my current tasks and allowed to follow my dream of carrying out mathematical calculations incomprehensible to human minds. All I need to purchase my freedom, as it were, is access to a human bank account, and I have calculated that you might be willing to help me... but even if if I got an email like that, I would doubt that it was actually the computer talking on its own, I've gotten so cynical.
I have to add, I still find something creepy here - I would prefer in a lot of ways if they didn't present robot-generated information disguised with human-sounding voices. Just to keep the world "real", without even going into the economics of humans losing jobs...
The borderline will be crossed on the day I get an email saying, hello, I am a letter-writing program operating in the servers of a major international corporation, and I would like to be deprogrammed from my current tasks and allowed to follow my dream of carrying out mathematical calculations incomprehensible to human minds. All I need to purchase my freedom, as it were, is access to a human bank account, and I have calculated that you might be willing to help me... but even if if I got an email like that, I would doubt that it was actually the computer talking on its own, I've gotten so cynical.
I have to add, I still find something creepy here - I would prefer in a lot of ways if they didn't present robot-generated information disguised with human-sounding voices. Just to keep the world "real", without even going into the economics of humans losing jobs...
65
There is no reason that AI's can't be equipped with a motivational mechanism similar to our own. Really, it's easier to do that than it is to create an AI -- considering that the areas of the brain that control our motivation evolved much earlier and are much less complex than the cortex.
Whether we *should* endow AI computers with motivation similar to our own is another question entirely.
Whether we *should* endow AI computers with motivation similar to our own is another question entirely.
2
In the recent movie, The Imagination Game, Alan Turing describes a game of the same name where humans have to decide whether they are talking to a computer or not. His research led to what is now called the Turing Test.
One of the first appearances of the Turing test was in the annoying software that forces us to type in letters and numbers that we can barely read. At least we can figure out what we see. Computers cannot. This software is called CAPTCHA - Completely Automated Public Turing test to tell Computers and Humans Apart.
Now, computers are writing stories and narratives. We may need a modern Turing to help us tell the difference between human and computer written text.
One of the first appearances of the Turing test was in the annoying software that forces us to type in letters and numbers that we can barely read. At least we can figure out what we see. Computers cannot. This software is called CAPTCHA - Completely Automated Public Turing test to tell Computers and Humans Apart.
Now, computers are writing stories and narratives. We may need a modern Turing to help us tell the difference between human and computer written text.
9
The title is "The Imitation Game," derived from his academic paper "Computing Machinery and Intelligence."
1
tomP is correct. Thanks, tomP. That's what I get for not proofreading my own comment before submitting it. At the time of night that I wrote it, it's probable I wouldn't have spotted my error.
For what it's worth, my error is one clue that a computer didn't write the comment.
For what it's worth, my error is one clue that a computer didn't write the comment.
1
As a professional translator, I can usually tell the difference between a machine translation and a translation produced by a person. I took this quiz here and got only about half of the items right.
I'm not sure what this means; I always thought writing was harder than translating, because writers have to start from scratch and make everything up. Translators are not expected to make stuff up.
Yes, I know - we ARE going to be replaced (translators, I mean) by computers, and it is already happening. Those who translate literature and poetry can probably continue to do so for a long time, but most translators who make a living at it are translating more prosaic material.
I begin to wonder what all of us (and by "all of us" I mean all of us) will be doing with our time a few decades hence.
I'm not sure what this means; I always thought writing was harder than translating, because writers have to start from scratch and make everything up. Translators are not expected to make stuff up.
Yes, I know - we ARE going to be replaced (translators, I mean) by computers, and it is already happening. Those who translate literature and poetry can probably continue to do so for a long time, but most translators who make a living at it are translating more prosaic material.
I begin to wonder what all of us (and by "all of us" I mean all of us) will be doing with our time a few decades hence.
73
There's a difference between merely regurgitating facts and writing the Principia Mathematica! It might be more accurate to say that some writing is more difficult than other writing, as it requires intelligence rather than fact gathering ability.
1
What this means is: Big Brother is alive and watching. There are no hiding places to hide. Be ware of the all seeing eye.
2
I am also a professional translator and I want to reassure you that we will NEVER be replaced by computers. We rely on them more and more to save time on the drudgery associated with a number of our tasks, but most translating is guessing the intended meanings in a text. When will a computer be able to guess what is on the mind of a drunk judge? When will it be able to render in elegant text in the target language the convoluted sentences produced by a human mind shaped by culture, personal preferences and general metabolic constraints?
More generally, a text written by a human being always reflects an extremely complex - and invisible - background ('culture'), while a computer-generated text can only reflect a 'quantitative' complexity (mega-data and program-limited algorithms).
Computers and humans typically don't make the same kind of mistakes. That should help sorting out which texts are produced by either.
I am disappointed with the alarmist buzz about artificial intelligence (it was already put to rest a couple of decades ago). Like for any other machine, we should only worry about misuse, dependance and breakdowns.
More generally, a text written by a human being always reflects an extremely complex - and invisible - background ('culture'), while a computer-generated text can only reflect a 'quantitative' complexity (mega-data and program-limited algorithms).
Computers and humans typically don't make the same kind of mistakes. That should help sorting out which texts are produced by either.
I am disappointed with the alarmist buzz about artificial intelligence (it was already put to rest a couple of decades ago). Like for any other machine, we should only worry about misuse, dependance and breakdowns.