Hi Claire, excellent article and really insightful facts on algorithm recruitment. I completely agree how you mentioned the role played by technology in simplifying the recruiter's work. The time consumed and resources wasted could now be well applied in searching right candidates with relevant hiring tools. Spire's tool is first of its kind and helping customers with contextual algorithmic hiring
This is a very interesting article. Perhaps distinguishing between different aspects in the process of bringing talent into your organisation is helpful here. For example I would distinguish between the processes of attracting talent, sourcing and sifting applications, recruiting and then selecting talent. It would seem to me that the algorithms discussed in this article could be most helpful in the sourcing and sifting stages of the process. Of course, as the article acknowledges, everything depends upon the quality of that algorithm, but if it can be unbiased the process of applying exactly the same criteria to everyone can only be an advantage. Sourcing seems to be an art as much as anything, and if this is a process which can access wider pools of talent, particularly in fields such as IT or engineering who permanently struggle with skills shortages, then it could help give a firm the edge in the war for talent. I would still advocate for the advantages of human interaction both during the recruitment and selection stages. I think if we can train our people to be involved at those stages with all the hallmarks of best practice then the value added from human interaction is still there. Interaction for applicants with personable recruiters and approachable interviewers acknowledges that the process is two-way, allowing the candidate to gain information and take a view on how they think they would fit in and I would be concerned about moves to eliminate that
Including talents that would be rejected by the subjective biased boss or colleagues does not fuarantee their integration. They will be rejected anyway sooner or later.
Mr. Anish Shah mentions Passion and Hustle. Passion is a word that is over used and should be given a decent burial. Anytime someone mentions passion my eyes glaze over. Hustle is great except when it produces a lot of Brownian motion and that wastes time and effort for a lot of co-workers.
However any hiring automation algorithm reduces some decision making power from inept HR people (not all of them, of course) and Managers who have unconscious bias is a good thing.
Now if only they could develop an algorithm to weed out toxic Managers and incompetent Managers who should not be managing people, productivity in American companies would skyrocket and there would be a lot of happy and satisfied employees. But I am asking for too much.
However any hiring automation algorithm reduces some decision making power from inept HR people (not all of them, of course) and Managers who have unconscious bias is a good thing.
Now if only they could develop an algorithm to weed out toxic Managers and incompetent Managers who should not be managing people, productivity in American companies would skyrocket and there would be a lot of happy and satisfied employees. But I am asking for too much.
1
The real deal is that algorithms can help humans hire better. The idea that algorithms hire better than a human misses the point. (Statistically, as others mention, valid pre-hire assessments are as good as or better than human judgment.) Good pre-hire assessments introduce speed and enhanced data quality into the decision process and faking the assessment is really hard (and obvious). But the assessment shouldn't take the place of a human decision maker. The assessment is like the musicians "blind audition" - managers can use the data without the distraction of human bias. The idea behind algorithm's is to match a job candidate's strengths to the requirements of the job. Strong match = better performance, lower turnover, reduced hiring costs. Now if the company has hiring managers who are poorly trained and/or are jerks, well... no algorithm is going to fix that.
3
Very interesting article with a somewhat silly title. Since the 1950's hundreds of studies have been done comparing algorithms to decision makers and the algorithms tie or beat the experts . . . always. Throw away the "John Henry" anecdotes and stop pretending that this question hasn't been answered hundreds of times. Algorithms are incredibly valuable and they are here to stay.
With that it mind it's also important to note that the scores that go into the algorithms are often based on expert judgment. Motivation, for example, can be measured by a test or by an interview question or using biographical data or (optimally) using a weighted combination of all valid predictors. Experts have a role to play and algorithms help experts combine the data points in an optimal fashion. In other words, as many commenters have noted, it's not an either or proposition. Great experts with great algorithms beat any other system.
I'm biased on this topic because my company has been building and validating algorithms for hiring since 2001 (http://www.stangdecisionsystems.com).
With that it mind it's also important to note that the scores that go into the algorithms are often based on expert judgment. Motivation, for example, can be measured by a test or by an interview question or using biographical data or (optimally) using a weighted combination of all valid predictors. Experts have a role to play and algorithms help experts combine the data points in an optimal fashion. In other words, as many commenters have noted, it's not an either or proposition. Great experts with great algorithms beat any other system.
I'm biased on this topic because my company has been building and validating algorithms for hiring since 2001 (http://www.stangdecisionsystems.com).
Great article, Claire. This article is very pertinent to the staffing industry and shows the real benefits of automated recruiting software. Our product, Leoforce software, is one of the companies pioneering this new technology. Thanks for sharing your insight!
I just went through a job search and many interviews, before accepting an offer.
Companies have taken rudeness during the interview process to a new level. No amount of data anlytics or algorithms is going to replace the fact that people - even those with advanced educations and in demand skills - are being treated like garbage and the reputation of companies that treat people this way is going to suffer.
I have been given ridiculously long assignments to compare as part of interviews, only to get generic HR rejection notes. I have been ignored completely and not given a response (too hard to email someone a no?) after going through multiple rounds. I have been put through 5 hours of interviews without being offered some water or a bathroom break.
The offer I ended up accepting - everyone was polite, HR did things according to the timeline that they clearly communicated, the interview was difficult but did not impinge on too much personal time or ask for free work and most important - clear, concise and friendly communication from start to finish. Do that, and you will attract better candidates that are not desperate to work for the first jerk company that hires them.
Companies have taken rudeness during the interview process to a new level. No amount of data anlytics or algorithms is going to replace the fact that people - even those with advanced educations and in demand skills - are being treated like garbage and the reputation of companies that treat people this way is going to suffer.
I have been given ridiculously long assignments to compare as part of interviews, only to get generic HR rejection notes. I have been ignored completely and not given a response (too hard to email someone a no?) after going through multiple rounds. I have been put through 5 hours of interviews without being offered some water or a bathroom break.
The offer I ended up accepting - everyone was polite, HR did things according to the timeline that they clearly communicated, the interview was difficult but did not impinge on too much personal time or ask for free work and most important - clear, concise and friendly communication from start to finish. Do that, and you will attract better candidates that are not desperate to work for the first jerk company that hires them.
6
Amen to this. I've experienced this sort of rudeness repeatedly. I have also seen the extent to which hiring no longer seems to be based on whether you can get along with people and do the job well, instead it's become a popularity contest.
1
My problem with these assessment tests is that they expect yes/ no, true/ false answers to questions best answered occasionally/ depends on circumstance/ depends on the goal. When they ask 'have you ever hated anyone', are they asking about a momentary feeling at a traffic light over the idiot who has 'parked' the intersection closed, nullifying the light? Or are they asking if I hold a lasting grudge hate against anyone? Or if there are historical figures whose historical record so disgusts me that I could say I hated them? When they ask 'Do you prefer the idea or the reality', well, it depends. The idea of traveling to the space station entrances me but I am pretty sure my version of claustrophobia (mine is more about not being able to escape at will) would make the reality a darned miserable experience. Even when applied solely to work environments, my response remains the same. I have had momentary feelings of hatred toward a fellow worker that had the net effect of reducing my overall respect and trust of that person but did not prevent me from being able to work on a project and produce the desired outcome. The five bubble answer probably provides more reliable data. But you still need a face-to-face interview and the algorithm might just make you miss interviewing the one person with the quirky imagination and logic leaps that keep a team on its toes and creating new concepts.
9
How do you promote your algorithmic app for (fill in the blank) hiring/ trading/ dating/ pretty-much-anything? You start by attacking… with some reductionist behavioral studies that demonstrate that humans are biased and cannot be trusted: you wear red ties? You'll only hire red-tied candidates. You throw in some other irrelevant studies too, as long as they're negative. Second, you give some far fetched examples that suggest that yes, you too, reader, could benefit from this app… you might get hired as a CEO of Apple on the basis of your school project back in 1987. And third, you add a political framing to the story and add that this will help minorities. Now, who knows? This may work. But there's a risk that people's professional experience may resemble the algo worlds it emulated of trading and dating. A company may not be able to hire a soul on a given year because its hiring algos crashed the process, just as HFT crashed the market in 2010. As for bringing to the hiring process the spirit of online dating, I can only speculate what joys that might bring.
11
This is always going to be hard evaluating the success of these programs. Just as it is hard to evaluate HR departments themselves. How do you decide how well the recommended person worked out? When the employee is not so good, is it because there own skills and abilities, or problems with co-workers, or because of a bad manager, or something else? And what about the essential control experiments - hiring by HR people, and hiring mostly at random once certain criteria are met? There has to be a systematic comparisons, but it would seem that there are too many uncontrolled variables that would make a sound comparison very hard. This is good for the companies writing the software, as they can make any sort of exaggerated claims without any objective evidence.
The saving grace might be that if hiring is mostly crap shoot, then you want the cheapest crap shoot, and the algorithms would provide that.
The saving grace might be that if hiring is mostly crap shoot, then you want the cheapest crap shoot, and the algorithms would provide that.
6
"The tech industry is a focus for some of the hiring start-ups in part because it has more jobs than it can fill...'
This is wrong! This is said only to convince legislators that more H-1B visa holders are needed. This digression is orthogonal to the topic, but, ironically, the software should pic the ideal candidates regardless of their layoff status, and sometimes decide the best candidate should get his/her job back!
This is wrong! This is said only to convince legislators that more H-1B visa holders are needed. This digression is orthogonal to the topic, but, ironically, the software should pic the ideal candidates regardless of their layoff status, and sometimes decide the best candidate should get his/her job back!
22
Amen to that! Anybody who says there aren't enough STEM people to fill the jobs hasn't been in the real world talking to unemployed and underemployed scientists and engineers.
3
It's funny how algorithms and computers are hiring people just as people are losing jobs to algorithms and computers
22
Hiring is so hit or miss, it would not surprise me. Do I think it could do a better job than me? Probably. I'm not that good at it.
3
Algorithms should complement, not substitute a person's decision.
10
Can an algorithm spot a liar? Can an algorithm spot someone who shows up for a job influence under the influence of drugs or alcohol? I've spent 30+ years in systems design & implementation which included interviewing candidates & designing software algorithms & the answer is definitely NO.
12
I would love to see HR out sourced by algorithms. Especially if the scripts are created and run offshore.
Good for shareholder value. Trickle Up in action.
Good for shareholder value. Trickle Up in action.
11
Perhaps this software could be applicable to the election process, where we the people have the computer select our leader based on his or her personal values, foreign policy views, morality, and other true human leadership qualities that aligns with our constitution's broad interpretation...
11
We have been using version of standardized test for years - some of which remind me of taking the SAT's they are so difficult for the executive group. While they give us some information I always take them with a grain of salt when they steer me on the "fit" section because the questions they ask are so general. There are some good indicators but I'll take behavioral interviewing and good reference checking every time...
7
As an engineer married to a headhunter, I have zero faith in algorithms. My wife has a Master's Degree in Industrial Organizational Psychology, the field that has help fuel this move. As an undergrad with a 4.0 she took the GMAT and did badly -- her background was NOT in business math. She was denied admission into the IO Psych program. She then studied for the GRE with a tutor and hit 80% and was then admitted. Because she missed round 1 all the psych assistanceships were gone. But her math score got her a position teaching math. So on the basis of ONE TEST she goes from being inadmissible to getting paid. By the way, her career has been stellar.
9
Why not? A match making by algorithm?
In places like India, historically match making were done by a crude version of algorithm called horoscope, which contained bits and pieces of data of potential brides and grooms.
A future ruled by algorithm is not too far.(Oh lord, please get me out of here)
In places like India, historically match making were done by a crude version of algorithm called horoscope, which contained bits and pieces of data of potential brides and grooms.
A future ruled by algorithm is not too far.(Oh lord, please get me out of here)
4
It would be harder to ignore women and older workers if algorithms play a significant role in hiring. It would be even harder to discriminate against very African American sounding names.
5
The multifaceted hiring process can be enhanced by the application of an algorithm just the hiring process is enhanced by competency exams and psychological tests. Technology has brought a vast array of "things" that are often coveted as the next better mousetrap. After nearly 4 decades in executive and professional talent acquisition and services I have seen the "Hiring process" move from rotary dial phones and classified newspaper advertisements to today's technologically fattened reality. If you want to buy a lawnmower or a computer, technology can expose all of a products pros & cons thus making a purchase decision easier & likely a better choice. We are speaking here about hiring human beings with all their faults and assets but mostly their unlimited potential and creativity. I strongly believe that we need to stop allowing technology to relieve us from our responsibility to make a great choice. When a "bad" hire occurs, it is easy to blame the "test". I have seen many incredibly qualified professionals passed over for hire due to a decision-makers lack of confidence to make a decision. It is people who create technology so let us pull up our proverbial socks and make a choice based on both techno data and our human instincts. Mankind's greatest accomplishments since the wheel were the product of human creativity and desire to improve things. Keep adding technology to the responsibility of choosing the best people and maybe you'll end up hiring a computer.
6
Political correctness has become the new golden calf. Hiring, just like working with humans is a subjective experience. We need to be able to"hang out" with others if we are to be productive. Political correctness can be easily faked and sensitive language cliches can easily fool a machine. In the end, it's another human that makes the algorithm for hiring me, my colleagues or my boss. I may not like to hang out with that person.
7
Most of the recommended comments here seem to poo-pooh the idea but I'm going to buck the trend and offer that an algorithm helped me find the man I married. Did the same for millions of others and no doubt produced millions more quality, if more temporary, romantic relationships. And if any part of the hiring algorithm approach reduces screener bias then bonus points.
The key is to keep the algorithm limited to sussing out things like minimum qualifications (Yes or No only, not how many years---too easy to ferret out age that way) and potential match between prospective employee work style and values with company culture and needs. If a candidate passes the screening process then let them take a skills test, if they pass that too then on to meeting people face-to-face for structured interviews. It would insure that companies don't select early for familiarity (or worse). Internal prospects and candidates recommended by colleagues should have to take the same skills tests so at least if they don't score well but still get the job the companies have to be honest with themselves about who they're hiring and why.
The key is to keep the algorithm limited to sussing out things like minimum qualifications (Yes or No only, not how many years---too easy to ferret out age that way) and potential match between prospective employee work style and values with company culture and needs. If a candidate passes the screening process then let them take a skills test, if they pass that too then on to meeting people face-to-face for structured interviews. It would insure that companies don't select early for familiarity (or worse). Internal prospects and candidates recommended by colleagues should have to take the same skills tests so at least if they don't score well but still get the job the companies have to be honest with themselves about who they're hiring and why.
4
I'd agree that all too often, hiring managers are looking to hire people that they'd like to hang out with and not necessarily those who would best do the job. On one hand that makes sense as they are looking to build camaraderie in the office, so a homogeneous group may be more likely to display that. This does of course lead to the exclusion of "the other" be they workers of color, women in male dominated fields, older workers etc. If the computer algorithms find a way around this it could open the doors to many who are excluded. Unless of course the parameters are set to exclude those over 45 or some such thing.
6
Having directly hired over 250 people during my business career, I always relied on the 'first born' test, that being don't hire them, they are always not team players, suffering from 'entitlement' and 'special' traits. Kids born in the middle always have better social skills.
2
I assume you are joking. How would you even know what an applicant's birth order is anyway. You're not allowed to ask personal questions like that in an interview, and I think most job applicants, unless they were desperate, would not take a job from someone who asked that question anyway.
12
First born won't try to outrank you, take your job, try to steal your wife, or complain while the organization dwindles into mediocrity, either.
4
Your comment seems to ooze "first-born entitlement". I agree with Jal's thoughts. The only possible way for you to assume an applicant's birth order without asking is to see your special traits and sense of entitlement in them. Are you afraid of hiring an individual who could replace you? To me, that would be one of my top three traits for hiring someone. It bodes ambition to succeed. Anything less will eventually become a wasted hire. True leaders and managers are not so ego-driven to assume that they are irreplaceable. Entitled individuals, on the other hand, often share the jaded trait of perfection. It's usually these people who eventually butt up to new senior leadership and loose out.
Total transparency of hiring/firing and promotion/demotion would eliminate much bias--legal and illegal--and result in more robust decisions. This means corporations would have to post for all to see (both within and outside the corporation) why and how these hidden HR decisions are made, and in this case how the algorithm works. Social pressure and the desire of corporations for social legitimacy would result in significantly less of what makes us all burn--unlawful and unfair discrimination based on race, gender, nepotism, cronyism, disability, and ethnicity. Allowing an individual or firm to fail as a result of not hiring on the merits is not enough. In that case, there would be no redress for the harmed, only ‘punishment’ maybe for the transgressor.
Totally transparent decisions would leverage opinion and the consequences thereof. This would be a move to more of a meritocracy. Firms would benefit as well by having the best-qualified individuals in jobs. Less of 'who you know' and more of 'what you can do.'
Firms would cry confidentiality of corporate secrets and intellectual property, and individuals would act for protection of individual privacy. There is a tradeoff between transparency enabling a vision of fairness and opportunity for all, compared to corporate confidentiality, and individual privacy. What is more important?
Let's see that algorithm.
Totally transparent decisions would leverage opinion and the consequences thereof. This would be a move to more of a meritocracy. Firms would benefit as well by having the best-qualified individuals in jobs. Less of 'who you know' and more of 'what you can do.'
Firms would cry confidentiality of corporate secrets and intellectual property, and individuals would act for protection of individual privacy. There is a tradeoff between transparency enabling a vision of fairness and opportunity for all, compared to corporate confidentiality, and individual privacy. What is more important?
Let's see that algorithm.
6
Oh great, here come the robots again.
Yes, I want to see evidence-based decision making and elimination of irrelevant biases. But given the ruthless corporate drive for cost efficiencies, this implies another whole bunch of folks are out of a job.
Yes, I want to see evidence-based decision making and elimination of irrelevant biases. But given the ruthless corporate drive for cost efficiencies, this implies another whole bunch of folks are out of a job.
9
Look back to what our country has achieved over the past 150 years - most of that in a non-digital age. There is no way to tell if any employee is a good fit or not until that person has worked at their job for a few months at least. I've hired people who I thought were perfect, only to realize after a few months that they didn't belong in the company and I've hired people who were a second or third choice who ended up being great employees. Graduating from college in the late 70s when companies would hire and train almost anyone, I am dismayed by the hyper-evaluation and long list of requirements for college grads now. To all hiring personnel out there, stop over evaluating and take a chance on our youth. They made it through college, so give them a chance. And consider older workers. They have years of experience and wisdom that are priceless (and remember, you will be "old" one day soon . . . if you are lucky).
45
Ah, everything old is new again. As part of academia that actually does the science behind these algorithms, we know exactly how to build this so it works. We usually call it synthetic validity when selecting for the job or "person-organization fit," when selecting for the company. Early efforts started in the 1950s, but we made some critical breakthroughs about a decade ago:(http://www.sciencedaily.com/releases/2006/03/060308085903.htm)
There have been special issues on how to conduct it:
Steel, P., Johnson, J. W., Jeanneret, P. R., Scherbaum, C. A., Hoffman, C. C., & Foster, J. (2010). At sea with synthetic validity. Industrial and Organizational Psychology, 3(3), 371-383.
And the number of these new companies that rely on the science? To the best of my knowledge, zero (plus or minus one). Just because it is delivered over the Internet, doesn't make it novel or well designed. By the way, we could build it so adverse impact is down to a trickle as well. It is special world where the disdain of science is such that you are getting this news as a commentary to a blog.
There have been special issues on how to conduct it:
Steel, P., Johnson, J. W., Jeanneret, P. R., Scherbaum, C. A., Hoffman, C. C., & Foster, J. (2010). At sea with synthetic validity. Industrial and Organizational Psychology, 3(3), 371-383.
And the number of these new companies that rely on the science? To the best of my knowledge, zero (plus or minus one). Just because it is delivered over the Internet, doesn't make it novel or well designed. By the way, we could build it so adverse impact is down to a trickle as well. It is special world where the disdain of science is such that you are getting this news as a commentary to a blog.
4
The author neglected to write about one of the most successful of the recruiting software packages out there " Connectcubed" at Connectcubed.com. It employs sophisticated psychometric analysis and high level game methods to assess competence and fit for any type of employment. It has a high correlation with successfully performing candidates.
This approach to hiring, as well as recruitment into professional schools, and professional positions is definitely the future. A person to person interview is also an important part of the process. The latter's success will improve immensely with the data from the former on hand ,serving as the backdrop for the interview.
This approach to hiring, as well as recruitment into professional schools, and professional positions is definitely the future. A person to person interview is also an important part of the process. The latter's success will improve immensely with the data from the former on hand ,serving as the backdrop for the interview.
2
I think one person can do a job pretty much as well as another, especially a low-level job, or a production job. Hiring for higher level positions that require more experience and expertise means you may have to be a little more selective. But, on the whole, companies spend way too much time, money and energy in hiring. It's not rocket science. People we think are a great fit end up leaving in a year. Someone that looks perfectly healthy spends most of their time dealing with chronic illness. Or, everyone in the department hates someone you liked very much. On the other hand, someone that you think is a bit dull, or is older, or who has slightly imperfect experience can turn out to be one of the best employees you've ever had. Hiring is a crapshoot. I don't see how an algorithm can change that. People do what they do, whether a computer program thinks they will or not.
39
Amen to that. In my experience, companies do spend excessive amounts of time making hiring decisions, or seem unwilling to make a decision at all. I think the reason is they're trying to cover their butts in case the hire doesn't work out.
Will the hiring company or the software developer be held accountable for the ageism that is bound to be coded into the algorithm? Whereas a human bias can be overcome when face to face with a prospective employee through conversation etc.., the bias in the algorithm will have the final say by not even considering an applicant that may fall above a predefined age limit. Algorithms and Big Data are not necessarily the solution to making every aspect of life and business better. Human intuition is going to be hard to replicate or replace.
11
This might be a great solution if the criteria that created the algorithm were made public as that might help eliminate points that are preventing many from getting jobs. Unfortunately criteria that have prevented me from getting a job ever since the financial crisis such as being over 50, being out of work for a long time, and having a poor credit score, can be incorporated into an algorithms just as easily as they can be stuck on a post-it on a recruiters computer.
26
The two non-ageist criteria are, of course, factors that might make a highly motivated employee who really appreciates having the job. And older employees don't have to learn to be responsible team players—about the time you hit 50, you realize you aren't the center of the universe. These three criteria are extraneous to job performance, and employers who use them—via algorithms or not—are part of that punitive "be sure to kick 'em while they're down" mindset that's soured our society. The two non-ageist criteria are part of the ground-level ways inequality is perpetuated or deepened, because they shift opportunity away from those who most need it and toward those already enjoying more security. If you were trying deliberately to create a permanent underclass, enforcing those two winnowing criteria would be effective tools.
1
Choose your computerized algorithm wisely, or it soon will be nicknamed GIGO.
4
I have seen terrible interviews given by employees and managers, so it bodes well for an expert system to determine qualified candidates.
One advantage is that a computerized system can handle vast amounts of data; analyzing candidate data for it is a piece of cake. A human can't analyze all of those factors, albeit they be small in number.
But there is the very real danger that hiring managers will bypass his or her people in favor of the system, and unquestioningly trust the computer. How many of us have tried to correct our data with an emplouse that tells us, "Well, that's not what the computer says!" If managers aren't capable, a computer is not going to make them capable.
The best way to design these systems is to make their decisions transparent to their users, and, indeed, to make no decisions whatsoever. Simply organize a candidate's capabilities. Right, maybe a candidate has no degree, but an analysis of his or her experience, against what people *really* do in a company, could be positive.
But this means that the expert system needs to know the business strategy of a company in order to increase its accuracy and make human-type analyses of its candidates. Do companies allow these systems to have access to thieir proprietary information?
One advantage is that a computerized system can handle vast amounts of data; analyzing candidate data for it is a piece of cake. A human can't analyze all of those factors, albeit they be small in number.
But there is the very real danger that hiring managers will bypass his or her people in favor of the system, and unquestioningly trust the computer. How many of us have tried to correct our data with an emplouse that tells us, "Well, that's not what the computer says!" If managers aren't capable, a computer is not going to make them capable.
The best way to design these systems is to make their decisions transparent to their users, and, indeed, to make no decisions whatsoever. Simply organize a candidate's capabilities. Right, maybe a candidate has no degree, but an analysis of his or her experience, against what people *really* do in a company, could be positive.
But this means that the expert system needs to know the business strategy of a company in order to increase its accuracy and make human-type analyses of its candidates. Do companies allow these systems to have access to thieir proprietary information?
2
Don't forget, even though an algoritm is a complex mathematical tool, it is basically designed by (biased) human beings. Meaning, someone must define the criteria for selection, Someone must decide which candidates will be "in" and which are "out" of the process, which variable get priorities, etc. The algorithm just churns the numbers. But how do we know which candidates are "good" in the first place? This is where there are some questions.
16
Interesting article. It is clear that automation is entering the recruiting and hiring process, and that is a good thing, overall, as biases in hiring are well-documented.
However, not all automated recruiting models or online labor platforms are the same, and some rely on the candidates selecting the job, and not an algorithm to pick them. Liberation Medical ( https://liberationmedical.com/ ) is an example of this. This platform allows medical professionals and healthcare facilities to engage, negotiate and get paid on medical services contracts directly, avoiding costly and time-consuming staffing agencies. These staffing agencies take from 30-50% of the medical professional's wages and often charge the healthcare facility additional fees as well. By cutting out this greedy middle-man, the medical professionals can charge the healthcare facilities less than the staffing agencies, yet still earn more. This allows medical professionals to choose the job that works for them, at the rate that they are happy with. That means greater access to medical care for everyone, and a more fluid and fair labor market for medical professionals.
Additionally, the Liberation Medical platform addresses the well-documented gender pay gap by allowing users to remain anonymous when proposing their rates of pay. This matters, as a male physician and a female physician (or nurse or therapist) with the same education, certifications and experience should receive the same wages.
However, not all automated recruiting models or online labor platforms are the same, and some rely on the candidates selecting the job, and not an algorithm to pick them. Liberation Medical ( https://liberationmedical.com/ ) is an example of this. This platform allows medical professionals and healthcare facilities to engage, negotiate and get paid on medical services contracts directly, avoiding costly and time-consuming staffing agencies. These staffing agencies take from 30-50% of the medical professional's wages and often charge the healthcare facility additional fees as well. By cutting out this greedy middle-man, the medical professionals can charge the healthcare facilities less than the staffing agencies, yet still earn more. This allows medical professionals to choose the job that works for them, at the rate that they are happy with. That means greater access to medical care for everyone, and a more fluid and fair labor market for medical professionals.
Additionally, the Liberation Medical platform addresses the well-documented gender pay gap by allowing users to remain anonymous when proposing their rates of pay. This matters, as a male physician and a female physician (or nurse or therapist) with the same education, certifications and experience should receive the same wages.
6
As as society I think we have been dehumanized more than enough. To strive for technical excellence (aka "success") at the expense of all else results in more sterile relationships. At the root is a lack or satisfaction with "good enough." (Another helping of greed or low self-esteem, anyone?) Hiring managers would be able to make more of a case that the bad hire "wasn't my fault. Talk to the algorithm." There would be less perceived need to manage and nurture a growing relationship with employees, and more reason to avoid taking personal responsibility (as if we weren't already doomed in that area already). It's more comfortable to manage employees and coworkers as productivity machines (a project component that sometimes hurts our feelings, doesn't do what they're told, doesn't understand, or works too slowly).
Fortunately, since we are dealing with messy, imperfect human beings which we are still not close to being able to quantify (have an algorithm yet to predict your spouse's mood tomorrow?), we will need messy, fuzzy algorithms that are as mysterious as we are--another tool, but still just a tool.
Fortunately, since we are dealing with messy, imperfect human beings which we are still not close to being able to quantify (have an algorithm yet to predict your spouse's mood tomorrow?), we will need messy, fuzzy algorithms that are as mysterious as we are--another tool, but still just a tool.
7
Ever since the "Employment Department" morphed into "Human Resources," employees have been seen as a consumable commodity rather than an asset to a company's growth and development.
Employers who value employees nurture loyalty and dedication among them: seeing them as algorithms instead of humans with unique attributes is a sure sign of bankrupt management.
Employers who value employees nurture loyalty and dedication among them: seeing them as algorithms instead of humans with unique attributes is a sure sign of bankrupt management.
16
Through out the civilization process and industrialization in last 200 odd years we have been trying to overcome shortcomings or deficiencies of humans.
The machines / processes / mechanisms we have invented or designed primarily do things what Humans can not do or can do better than Humans.
Algorithm when prepared considering minute details and validated from time to time, will surely be able to take consistently better hiring decisions.
Objective? Some machines / inventions make things faster, some safer, some less stress generating for Humans. This will help Humans reduce stress of decision making (hiring) process along with avoiding blame of wrong hiring.
The machines / processes / mechanisms we have invented or designed primarily do things what Humans can not do or can do better than Humans.
Algorithm when prepared considering minute details and validated from time to time, will surely be able to take consistently better hiring decisions.
Objective? Some machines / inventions make things faster, some safer, some less stress generating for Humans. This will help Humans reduce stress of decision making (hiring) process along with avoiding blame of wrong hiring.
1
It seems to me that this will simply result in a cottage industry of resume consultants who try to figure out a way to beat the algorithm. The disadvantaged, who don't have the financial resources to access such consultants will still be out in the cold.
Hiring will always to some degree be subjective and imperfect. Just because some computer algorithm comes along does not necessarily change that; those who write the algorithms could have biases of their own. Moreover, there's an additional problem with algorithms. That problem is that sometimes you hire someone who's not the perfect fit for a given job simply because you think they're going to have the ability to move up into more advanced positions with greater responsibility. I'm not sure how you write an algorithm for that. Algorithms can measure where someone is, but aren't likely to do well at predicting where they could go.
Hiring will always to some degree be subjective and imperfect. Just because some computer algorithm comes along does not necessarily change that; those who write the algorithms could have biases of their own. Moreover, there's an additional problem with algorithms. That problem is that sometimes you hire someone who's not the perfect fit for a given job simply because you think they're going to have the ability to move up into more advanced positions with greater responsibility. I'm not sure how you write an algorithm for that. Algorithms can measure where someone is, but aren't likely to do well at predicting where they could go.
17
Don't forget that you can program in bias deliberately.
10
"One engineer had applied twice to Rackspace, a cloud computing company, without luck. As an Army veteran who worked in public radio with no high school degree or professional programming experience, he did not fit the pattern that Rackspace looked for. But Gild suggested him based on the software he had been writing on his own, and he was hired."
It would have been helpful to learn whether or not this was successful. Did he/she work out or not? Without knowing that, this anecdote is not evidence for (or against) the superiority of algorithmic hiring.
It would have been helpful to learn whether or not this was successful. Did he/she work out or not? Without knowing that, this anecdote is not evidence for (or against) the superiority of algorithmic hiring.
12
Hiring by algorithm? It's already happening, in the form of using online personality tests to weed out job applicants who don't fit every characteristic The Boss desires. Not too surprisingly there are lots of people who can't even get an interview, as well as a lot of (in)human resource types who constantly bemoan the fact that they just, in the tones of the endless whining of so many Palm Beachers, 'can't get good help'. Really.
Of course, the real problem is simple: People don't work that way. Perhaps the ultimate, futuristic dream of these algo-duped employers is a sort of 'Stepford-Meets-R.U.R'. scenario. For those enamoured of that idea, I suggest that they pick up a copy of Karel Capek's 'R.U.R.' and actually read the play, all the way through. Hint: It doesn't exactly end well.
Of course, the real problem is simple: People don't work that way. Perhaps the ultimate, futuristic dream of these algo-duped employers is a sort of 'Stepford-Meets-R.U.R'. scenario. For those enamoured of that idea, I suggest that they pick up a copy of Karel Capek's 'R.U.R.' and actually read the play, all the way through. Hint: It doesn't exactly end well.
12
The rule is this:
If you have a group of candidates with the same qualifications & they match the jobs specs, hire any one of them at random & they'll fit.
All the rest of it is a bunch of people earning a living by pandering to insecure &/or inexperienced &/or dimwit 'managers'.
I could write a book about the cretinous interviews I've been to. This includes shills in the so-called 'recruitment industry' who had no idea what any of the IT words they were asking about actually meant.
The more you pay to hire staff, the more likely you are to praise & defend that process.
If you have a group of candidates with the same qualifications & they match the jobs specs, hire any one of them at random & they'll fit.
All the rest of it is a bunch of people earning a living by pandering to insecure &/or inexperienced &/or dimwit 'managers'.
I could write a book about the cretinous interviews I've been to. This includes shills in the so-called 'recruitment industry' who had no idea what any of the IT words they were asking about actually meant.
The more you pay to hire staff, the more likely you are to praise & defend that process.
8
Companies want idiot savants. From my experience, companies want workers that are stupid enough to take advantage of and yet smart enough to do the job.
Computer algorithms probably can detect idiot savants more efficiently than the idiot savants that run HR.
Computer algorithms probably can detect idiot savants more efficiently than the idiot savants that run HR.
23
Amongst biases and imperfections, how does the algorithm forecast to deal with close matches and don't you think that keyword-based matching is not an equivalent for human intelligence as in rational thought?
What algorithms are going to be designed to create abstractive machine learning and is this going to be good enough for job candidates and the employer?
Also, how do we enable a monitoring process that is independent of the machine?
I do not believe we can do away with the human frankly.
What algorithms are going to be designed to create abstractive machine learning and is this going to be good enough for job candidates and the employer?
Also, how do we enable a monitoring process that is independent of the machine?
I do not believe we can do away with the human frankly.
I think it is more about what kind of people are taking up HRM as a profession. Are they aware of implicit biases they have?? Is the organization aware of implicit bias as a disadvantage?? It is as much important to have a system(Automated or not) to check biases and prejudices. Organizations must in the wake of growing inevitability of talent management,monitor for fatal biases in HRs. When the CHRO has a strong value system and is treated more as a strategic partner, selection and recruiting become more refined and calibrated decisions are made.
Besides, automating the hiring process can turn itself into the same faulty system that these start ups are talking about. They are trying to "Fit in" certain employees with certain attributes to specific teams and jobs. How different is that from prejudiced selection process?? It does not help the system change but makes do with the prevailing on.
Ideally, such start ups must work in collaboration with organizations to help formulate best practices in HR which exterminate biases.
Companies such as Textio can be game changers in HR analytics.
Besides, automating the hiring process can turn itself into the same faulty system that these start ups are talking about. They are trying to "Fit in" certain employees with certain attributes to specific teams and jobs. How different is that from prejudiced selection process?? It does not help the system change but makes do with the prevailing on.
Ideally, such start ups must work in collaboration with organizations to help formulate best practices in HR which exterminate biases.
Companies such as Textio can be game changers in HR analytics.
1
“It’s an intuition, gut feel, chemistry.” He compared it to first meeting his wife."
How appropriate that Mr. Shah compared hiring to mate selection; both processes are rife with confirmation bias and, let's face it, both are often selected from groups of competent individuals who will, most-likely, perform well and appear to have been destiny five years down the road, in retrospect. In fact, employees will typically look better than that mates because few people dwell on the "what-ifs" of hiring like they do the "what-ifs" of marriage.
I think the algorithms will do just as well as the humans because, try as we may, we will not be able to resist setting the inclusion parameters. An algorithm doesn't equal artificial intelligence but, as people are hired and perform reasonably well, the algorithm will look like a miraculous solution when the software developers tell us how they have found the perfect algorithm to balance talent, "partnership" and "passion."
I can only wonder if the salaries of the laid-off HR department will offset the annual licensing fees.
How appropriate that Mr. Shah compared hiring to mate selection; both processes are rife with confirmation bias and, let's face it, both are often selected from groups of competent individuals who will, most-likely, perform well and appear to have been destiny five years down the road, in retrospect. In fact, employees will typically look better than that mates because few people dwell on the "what-ifs" of hiring like they do the "what-ifs" of marriage.
I think the algorithms will do just as well as the humans because, try as we may, we will not be able to resist setting the inclusion parameters. An algorithm doesn't equal artificial intelligence but, as people are hired and perform reasonably well, the algorithm will look like a miraculous solution when the software developers tell us how they have found the perfect algorithm to balance talent, "partnership" and "passion."
I can only wonder if the salaries of the laid-off HR department will offset the annual licensing fees.
6
I've been a professional recruiter for nearly 35 years for US and foreign companies in advanced high technology and other fields.
In short, the answer to the question asked in the title of the article is "no".
For decades many services, products, systems, methods, trainings, and the like have been offered to try to improve hiring results and they all promise pretty much the same thing.
None of them work. None have ever worked. I doubt, very seriously, that any ever will work.
Good hiring is an extremely complex and difficult process with many variables. Technical or professional skills and aptitude can be measured along with degree completion or training, and even intelligence. What cannot be measured except by people is compatibility with a company and job, empathy, drive, honesty, presence, behavior, bias, personal integrity, teamwork.....the list goes on and on and on and there are less than 550 characters left in this posting.
Don't bet the ranch.
In short, the answer to the question asked in the title of the article is "no".
For decades many services, products, systems, methods, trainings, and the like have been offered to try to improve hiring results and they all promise pretty much the same thing.
None of them work. None have ever worked. I doubt, very seriously, that any ever will work.
Good hiring is an extremely complex and difficult process with many variables. Technical or professional skills and aptitude can be measured along with degree completion or training, and even intelligence. What cannot be measured except by people is compatibility with a company and job, empathy, drive, honesty, presence, behavior, bias, personal integrity, teamwork.....the list goes on and on and on and there are less than 550 characters left in this posting.
Don't bet the ranch.
23
The effectiveness of algorithms is easy to prove. If, as their apologists say, these are better than humans, surely they do not fear sunlight and transparency. Each candidate, hired or not, should receive its score and the reasons of hire, or not.
9
Sometimes it seems that those who promote algorithms will say anything. They seem to have one reason or another that these are better. Of course they are usually selling their algorithm or service and it's easy to talk it up, the technology of it all.
But how do they know algoritms are really better? I haven't seen much convinvig proof. But I think they get cookie-cutter applicants, or people who know how to "game" the system with key word. Things like "work-related values", how do you get that in an algorithm? You get pop phrases.
But how do they know algoritms are really better? I haven't seen much convinvig proof. But I think they get cookie-cutter applicants, or people who know how to "game" the system with key word. Things like "work-related values", how do you get that in an algorithm? You get pop phrases.
12
The good news; hiring algorithms don't use hunches or intuition. The bad news: hiring algorithms don't use hunches or intuition. Managers at decent sized companies should be required to read "The Human Use of Human Beings" by Norbert Wiener. The book all programmers and IT management should be forced to read is "The Psychology of Computer Programming" by Gerald Weinberg. Hiring algorithms should or can be a component of selecting candidates for jobs, if the algorithm can be adjusted by senior management to incorporate its company identity and goals by using parametric choices i.e. soft coding, then the program will add a uniform unbiased review to all incoming hiring options.
Here is the bottom line, you can use a psychological inventory test, have HR interview the candidates, use input from a hiring algorithm and whatever else an HR consultant can sell you. Nothing will or should replace a manager's gut feeling created from job related knowledge, personal experience and wisdom, combined with having the best interests of the department and the company in mind.
The hiring algorithm is a tool. Anyone who places the balance of strategic power in an automated tactical tool will fail.
Here is the bottom line, you can use a psychological inventory test, have HR interview the candidates, use input from a hiring algorithm and whatever else an HR consultant can sell you. Nothing will or should replace a manager's gut feeling created from job related knowledge, personal experience and wisdom, combined with having the best interests of the department and the company in mind.
The hiring algorithm is a tool. Anyone who places the balance of strategic power in an automated tactical tool will fail.
5
Back in the 90s, when I was a C++ programmer, some of the top guys came up with a C++ test we gave to all job applicants. A surprising number of the candidates were unable to answer any of the 20 questions. A few of them answered all, or nearly all, of the questions correctly. We interviewed those guys to see if they were OK personally and would fit into the team.
Personal fit is very important. There are hard-core places where everybody works day and night, and only the toughest survive. If you have to attend lots of stupid meetings and still code and test twelve hours a day, most of the applicants won't be very successful in the job.
Personal fit is very important. There are hard-core places where everybody works day and night, and only the toughest survive. If you have to attend lots of stupid meetings and still code and test twelve hours a day, most of the applicants won't be very successful in the job.
4
" you’re also getting into discrimination. ”
Umm, the hiring process is an exemplar of discrimination from end-to-end and automating it makes that quite transparent. It is amazing how many in HR need to have that turned into Braille.
The discrimination that has been turned into a code-word will resurface at the last office the prospective hire goes into when a manager discovers there are other applicants that are better suited to the position.
The welcoming of that prospective hire won't feel the handclasp of a computer and that software isn't going to sign the new employee's check. As HR is always that last office, it will always remain to serve the employer, not ethics.
The other wheeze about "culture" is misappropriation of language there, too:
“A cultural fit is an individual whose work-related values and style of work support the business strategy,”
No, that is a definition suited to a Dilbertian sweat shop where culture can be drained to such banalities.
Umm, the hiring process is an exemplar of discrimination from end-to-end and automating it makes that quite transparent. It is amazing how many in HR need to have that turned into Braille.
The discrimination that has been turned into a code-word will resurface at the last office the prospective hire goes into when a manager discovers there are other applicants that are better suited to the position.
The welcoming of that prospective hire won't feel the handclasp of a computer and that software isn't going to sign the new employee's check. As HR is always that last office, it will always remain to serve the employer, not ethics.
The other wheeze about "culture" is misappropriation of language there, too:
“A cultural fit is an individual whose work-related values and style of work support the business strategy,”
No, that is a definition suited to a Dilbertian sweat shop where culture can be drained to such banalities.
13
Hiring algorithms may be brilliant, but what a spectacular waste of design talent. Tell me please, which financial advisor is crooked, which lover a cheater, which recidivist criminal has really and truly turned over a new leaf, which politician is lying. (Okay, maybe the last one is too easy because all are lying.) It pains me to think all of this mathematical effort is wasted on finding an assistant accounts manager for a pizza chain.
23
I am a web developer with multiple disabilities, I once went to a job interview where they gave me a quiz on web development which I aced. They didn't hire me--in fact they left the position open. That being said the culture in a lot of companies is not accommodating for anyone a step or more away from privilege so even if algorithms started hiring superstar candidates they still might be cast offs (Ben Folds' Fred Jones Part 2 comes to mind here).
4
Wow. So I assume we would all be working next to "Algorithms" instead of PEOPLE. Sure you can screen resumes using algorithms. You can predict weather using algorithms (yea and you see how accurate those predictions are) and you can write algorithms to do all sorts of other things. However, algorithms can measure a person's desire or their "heart". These intangibles also don't show up well on a resume. Many times what's on a resume is inflated or a flat out lie. Train people how to communicate better with people to find out their desires and motivations. Then try to match those to the right careers and work environments and you will have better success than we currently see in the hiring process. On the other hand, many companies now view people as replaceable assets and faceless expenses on the P&L statement. With that attitude towards employees why not use algorithms to hire them (and fire them when the company needs to downsize again).
2
Having an inside contact who can get you in front of the hiring manager is ALWAYS the best. With all the resume screening programs and data driven methodologies that are now widely used for recruiting, please tell me WHAT do we even need HR for? Payroll and benefits are often self administered by the employee thru the company website or outsourced completely. HR reps just add a layer of confusion to the job search process.
7
tell me WHAT do we even need HR for?
------------------------------
HR in a modern corporation is a profit center. Their job is to find cheats, tricks to squeeze labor cost. Four examples:
1. Collude with HR from competitive companies to lower industry wide wages - "we pay competitive salaries".
2. Pay 401k at the end of the year to effectively lower company contribution.
3. Create scheme to fire American workers and hire H-1B visa holders.
4. Export jobs below the executive suite to third-world countries.
------------------------------
HR in a modern corporation is a profit center. Their job is to find cheats, tricks to squeeze labor cost. Four examples:
1. Collude with HR from competitive companies to lower industry wide wages - "we pay competitive salaries".
2. Pay 401k at the end of the year to effectively lower company contribution.
3. Create scheme to fire American workers and hire H-1B visa holders.
4. Export jobs below the executive suite to third-world countries.
Computers are good at routine, humans are better at judgement. Smart hiring personnel will use both.
10
HR Department aside, you really need an inside contact, with some standing in a company prior to seeking employment. This prevents you from being at the mercy of the HR dept
7
This is what hiring has become, but it shouldn't come down to "who you know".
Sounds like a better and cheaper way to pick a President.....
6
The only algorithm I'm interested in developing is the one which disposes of the entire HR department.
61
Could not agree more! HR deptartments are totally useless and hopefully will be obsolete very soon.
5
So...you want to work for a company where there is no one to handle discrimination claims, organize training and development programs, manage conflict between managers and subordinates, ensure that compliance standards are implemented correctly, support business lines that need to hire new people, coordinate benefits plans, track compensation to ensure accuracy and market-driven raises, etc.? Sounds like a great place to work!
2
there is no other faster way to find the creme of the crop in software developers
1
To any employers hoping that an algorithm will find a them a unicorn,.... there has never been or will never will be an algorithm that will find them that unicorn,... unicorns don't exist.
HR departments, bad executives, bad managers, bad companies, bad owners, etc. will never go away. If enough anonymous data is collected, I hope that someday the algorithms do end up helping the labor market,.... and in turn, business.
HR departments, bad executives, bad managers, bad companies, bad owners, etc. will never go away. If enough anonymous data is collected, I hope that someday the algorithms do end up helping the labor market,.... and in turn, business.
7
considering how poorly HR departments perform, a chimpanzee can probably do better.
31
No, an algorithm cannot hire better than a human anymore than it could draft a better football team. There are a lot of intangibles. And yes, personality and looks count where employee is in sales or deals with the public.
11
Having been in a position a number of times to screen applications, the first and best thing to do with the application or resume is for the person who is opening or printing the applications to remove the name, sex, race or age of the applicant. Then start narrowing down the pile. If the algorithms start with that premise, they may be useful. As to HR offices in large corporations and government, my experience as a scientist is they are the most useless employee on the planet who fill an FTE that could be better used elsewhere. When you need to hire someone with a particular skill set, 90% of the time the HR office is clueless as to the qualifications for the new hire. Granted, it is a problem with scientific and engineering jargon, but even trying to explain it to an HR supervisor is like talking to a guppy. The totally blank expression on their face gives them away. If have a bias, it is towards competence in the employee.
49
I have lived through both eras-- the more HR is involved, the less relevant the conversations, as I am filtered for, apparently, not knowing buzz words. The conversations used to be specifically about the job, and the challenges, in technical language that the HR filterers I've encountered recently are not able to discuss in an in-depth problem solving way. So instead I am asked for an adjective about myself. One time my "problem solving" ability was tested by someone asking me how I would increase sales of chocolate chip cookies in a supermarket, instead of being asked about actually facing the organization and/or relevant to the areas of my training and planned focus. So I suppose we are supposed to study up on HR manuals instead of spending the same time gaining additional job-relevant expertise.
The point is: for companies that filter applicants for positions involving advanced math, engineering and science knowledge and experience through people uninterested in those subjects, can the algorithms be worse than what we have now?
The point is: for companies that filter applicants for positions involving advanced math, engineering and science knowledge and experience through people uninterested in those subjects, can the algorithms be worse than what we have now?
31
Given the American propensity for outsourcing, trying to find someone that's an exact match, and underpaying I'm not sure that these tools would help. On the other hand, they can't be much worse than some interviewers I've dealt with. I've been to interviews where the company was incredibly inconsiderate, where I was asked inappropriate questions, told that I was lying about my credentials and experience, or outright lied to about the job. I've had to take psychological tests to see if I fit the culture of the company rather than being asked what I could do, had done, etc.
My feeling about corporate America is that the skills we have don't matter. It's if we make our potential bosses feel good or don't come across as competent enough to threaten their positions. We're expected to be happy little campers that say yes all the time. Yes, step on me. Yes, make me work late even when it's not necessary. Yes, I don't mind coming in when I'm sick, working on my vacation time, sacrificing my whole life until you decide I'm too expensive or useless.
The bias seems to be towards men who can talk about baseball, football, hockey, beer, how the girls dress. I've been working over 30 years and it hasn't changed. Most job descriptions are inaccurate and most interviewers don't know how to interview. Most employers don't want to train or retain. And women are underpaid and undervalued.
My feeling about corporate America is that the skills we have don't matter. It's if we make our potential bosses feel good or don't come across as competent enough to threaten their positions. We're expected to be happy little campers that say yes all the time. Yes, step on me. Yes, make me work late even when it's not necessary. Yes, I don't mind coming in when I'm sick, working on my vacation time, sacrificing my whole life until you decide I'm too expensive or useless.
The bias seems to be towards men who can talk about baseball, football, hockey, beer, how the girls dress. I've been working over 30 years and it hasn't changed. Most job descriptions are inaccurate and most interviewers don't know how to interview. Most employers don't want to train or retain. And women are underpaid and undervalued.
64
Thanks for writing how it is and what I wanted to say. It is so very disappointing that with the newer generations, this has not changes, only to get worse. It is a discouraging world out there.
9
My feelings exactly. Companies don't want sharp go-getters, even when they say they do. What they actually want are doormats who do as they're told, never challenge or question a process or decision, and who are content stay in the same role for years. Since that is not me I tend not to do well in corporate America.
14
I'm still struggling with the realization that it's not at all how smart you are, how hard you work or how well you do your job; it's whether you fit in with the office culture that seems to predict how valued and well treated you will be in your job. Suck up to the boss, be one of the gang and you'll do just fine.
6
Someone has to program the algorithm, don't they? What's to keep their biases from creeping in? Eventually people will figure out how to write resumes that are more likely to be selected by the algorithm.
While the article cites the case of a worker at Rackspace who was different and did not match the conventional wisdom, I am more inclined to think an automated algorithm would work the opposite way and not detect potential employees who deviate from the norm.
While the article cites the case of a worker at Rackspace who was different and did not match the conventional wisdom, I am more inclined to think an automated algorithm would work the opposite way and not detect potential employees who deviate from the norm.
24
Part of the problem is that people are not conscious of their biases - they believe they are selecting on legitimate criteria, but in fact their decisions are based on feelings toward applicants that have nothing to do with their ability to perform.
Often the actual criteria that they'd like to think they're using to make hiring decisions are just fine; it's that these criteria are not actually the deciding factors in practice. In that case, programming the criteria you want into the algorithm and having it take care of deciding whether someone actually meets those criteria solves the problem.
Often the actual criteria that they'd like to think they're using to make hiring decisions are just fine; it's that these criteria are not actually the deciding factors in practice. In that case, programming the criteria you want into the algorithm and having it take care of deciding whether someone actually meets those criteria solves the problem.
5
Never mind algorithms, it's the requirements that are the problem. Filtered thru the HR dept., the basic needs of the department doing the hiring get translated into impossible, or at best unlikely, sets of requirements that can be easily parroted by an ignorant drone who gets first look at every application. The algorithm will be of no help there, it will just make the same mistakes as the human.
13
Great insight! Candidate X want to be CEO. X would make a great CEO, but has never held that positioni. One of the 'requirements' is past experience as a CEO. X will be screened out ... algorithm or no algorithm. But, how is past experience as CEO in any way a requirement for the job. It's not. It's just a way to incorporate bias... to ensure that any hires will already be members of the special CEO club. Now, if the algorithm could really say hire X to be CEO ...
1
Perfect, just what American business culture needs, more MBAs with algorithms.
Nothing beats experience and human interaction like theory and automation.
Nothing beats experience and human interaction like theory and automation.
20
The basic attributes of a great employee are always the same-only the skills needed in the job differentiate.
4
No they are not. In sales and public relations, looks, personality, attitude, communication skills, and many more attributes are required and are not checkable in an algorithm.
5
The algorithm may increase diversity in 5 ways, but if at the same time, it reduces it in one or two, it will still run afoul of disparate impact rules.
Perhaps algorithms could be useful for screening resumes, which often come in by the hundreds or thousands. Current hiring practices depend largely on interviews, which have no relationship to an applicant's ability to actually do the job, only on how personable, confident or good at marketing themselves they are. And, as the article points out, many biases distort the results.
7
Software is used to screen resumes. The more your resume fits in with the key words set up by HR the better your chance of getting an interview. The problem is that the software, no matter how good it is, cannot make up for poor programming. GIGO is the technical term. Furthermore, in our extroverted society, people who are shy, quiet, or don't brag about themselves tend to lose in the interview games.
23
kr, you are definitely wrong on the 'smart algorithm' situation. Any company that recruits or accepts applications online already uses "keyword scanning" and other non-human techniques to screen applicants. Only those that pass that gate will get looked at by HR or (heaven forbid!) a hiring manager.
One failure point is when the company or the applicant use the wrong keyword, or a generic term (say, "scripting") instead of specific examples (perl, bash, python).
One failure point is when the company or the applicant use the wrong keyword, or a generic term (say, "scripting") instead of specific examples (perl, bash, python).
5
Exactly. On one hand, there is a paragraph saying this is not meant to substitute for skilled people making decisions and on the other they say it may help dispell issues such as racial basis or workplace "compatibility" that simply assues more of the same profiles are hired.
The simple truth is resume sorting systems have existed for a LONG time. Some of them have embedded testing as the "second level" and some don't. But in the end, if skilled people should be involved, what is the difference between this idea and Resumix? Really?
The simple truth is resume sorting systems have existed for a LONG time. Some of them have embedded testing as the "second level" and some don't. But in the end, if skilled people should be involved, what is the difference between this idea and Resumix? Really?
1
At our firm, www.Skillbridge.co, we use technology to match consultants with projects, in a similar way to how OKCupid has revolutionized online dating. Using a mix of proprietary and public data, we help firms hire Elite Business Freelancers for projects. Therefore, I respectfully disagree with those who say "it's an intuition, gut feel, chemistry" when hiring, as we have significant quantities of data that proves otherwise.
1
I agree. I've felt the "intuition" and the "gut" work against me in many an interview, to say nothing of age discrimination.
5
And how do you judge the success of the consultant that you placed versus the one that you did not place in the same opportunity?
It is a case of an Induction Proof Check for a binary tree T - which any good programmer already knows.
It is a case of an Induction Proof Check for a binary tree T - which any good programmer already knows.
3
This reminds me of the 300K self cleaning toilets that were supposed to revolutionize public bathrooms. It was then disovered that humans could clean a version at almost a tenth of the cost. The savings never added up.
My last hire had such a desire for her position that I overlooked any other possible problems in her resume. She has been amazing. Other people who looked great on paper failed miserably.
My last hire had such a desire for her position that I overlooked any other possible problems in her resume. She has been amazing. Other people who looked great on paper failed miserably.
40
Just my luck, the recruiter who looks beyond "possible problems" is in Spokane while I am in New York City. Wish I could interview with someone like you.
2
One issue is that resumes just haven't changed in the past 100 years, and so employers are using the same old tools to find employees. Once we can replace resumes with a more streamlined system, the algorithms become much cxlearer.
There are also a number of sites that focus on the job-seeker instead of the employer. Some are niche sites, like military.com that allows users to enter their military history and searches jobs. Another is Job-Set.com that uses a dating-site style layout to match people to jobs quantifiably based on skills, knowledge, and behavior, and allows users to upload their college transcript or military history, and builds them a skill profile sheet to give to their potential employer.
There are also a number of sites that focus on the job-seeker instead of the employer. Some are niche sites, like military.com that allows users to enter their military history and searches jobs. Another is Job-Set.com that uses a dating-site style layout to match people to jobs quantifiably based on skills, knowledge, and behavior, and allows users to upload their college transcript or military history, and builds them a skill profile sheet to give to their potential employer.
2
Job-Set won't be up and running for another two weeks. Looks like it might be another good tool in the job search toolkit.
Can an algorithm pick the best candidate for an individual position? With time and further sophistication, yes probably. But I think it would be very difficult to capture the team aspect that exists within firms. I don't see how an algorithm, designed by flawed and biased humans, can capture individual personality quirks and how those will interplay with others in the company. Schumpeter featured an interesting piece on corporate recruiting and hiring ("How to join the 1%", 16 May 2015), in which a recruiter compares hiring to “picking a team on the playground”. So, I do think there is still a human element necessary to judge how well that individual would fit in. The Schumpeter piece also describes the importance recruiters place on confidence, which seems like a very tough trait for a computer to grasp. Anything that breaks down barriers and inequality in the workplace is great, but human intuition cannot be underestimated.
6
I'm sure the first statement in the algorithms will be, No one over age 45 need apply.
86
those people are just too expensive. If they worked at same salary as a 21 year old, and could do everything just as quickly, this would not be a problem. But this group just feels to entitled to high wages that just price them out of the market.
Yoda:
I agree with kat. Perhaps you're a twenty-something who thinks that "older" applicants all feel entitled to higher wages. I hate to disabuse of that notion. Many people in my age group would love to work full time at a decent salary (not necessarily for "high wages"), but automated HR screening systems (algorithms are a more sophisticated version of the same) (wo)manned by clueless HR folks simply ignore them. Trust me, there's bias, no matter how many clever mousetraps you invent, against older applicants, don't fit the so-called mold of "culture" fit and "executive presence" as defined by HR paper pushers. I speak from experience of being told so, in politically correct verbiage of course, many times that I have lost count. Unless you're a really smart HR person (not too many unfortunately) who can analyze beyond the string of mechanical key words in selecting a resume, no of algorithm will eliminate ethnic, race or gender bias. The only consolation I've is that the very people, who think the world of themselves because they're smart with algorithms, will in their own turn be cast aside when someone even younger will find a better way to make the 40-something employees redundant in corporate America. My generation will somehow make it to the finish line, but I'm afraid I don't see that bright future for yours.
I agree with kat. Perhaps you're a twenty-something who thinks that "older" applicants all feel entitled to higher wages. I hate to disabuse of that notion. Many people in my age group would love to work full time at a decent salary (not necessarily for "high wages"), but automated HR screening systems (algorithms are a more sophisticated version of the same) (wo)manned by clueless HR folks simply ignore them. Trust me, there's bias, no matter how many clever mousetraps you invent, against older applicants, don't fit the so-called mold of "culture" fit and "executive presence" as defined by HR paper pushers. I speak from experience of being told so, in politically correct verbiage of course, many times that I have lost count. Unless you're a really smart HR person (not too many unfortunately) who can analyze beyond the string of mechanical key words in selecting a resume, no of algorithm will eliminate ethnic, race or gender bias. The only consolation I've is that the very people, who think the world of themselves because they're smart with algorithms, will in their own turn be cast aside when someone even younger will find a better way to make the 40-something employees redundant in corporate America. My generation will somehow make it to the finish line, but I'm afraid I don't see that bright future for yours.
10
Well, that one might follow "What is your expected salary?"
1
How long will it be before someone writes software that will help tune resumes to match the HR algorithms? While esearch like that done by Daniel Khaneman has shown that using objective criteria is a much better way to select a candidate for the job than using your gut. However 40 years of reading resumes have taught me to look more deeply into a person's credentials. Often an offbeat background or a clever solution to a particularly difficult problem show much more about a person's innate abilities than degrees and a well crafted list of the acronyms listed in the skills section. My experience in high tech is that five years after graduation a degree only indicates that you had the financial resources to get into college and graduate.
67
My husband's company was making HR decisions based on an algorithm. They laid off a statistician on the grounds that he did not have a degree in mathematics or statistics. However, the man had a Master's in psychology; to earn that degree, he had taken a lot of statistics. He had successfully worked with statistics for years. It is very difficult for unemployed people in their fifties to get rehired; the ignorance, willful or inadvertent, of the HR department in making this decision, caused a lot of pain.
97
HR is intellectually bankrupt. They are the most incompetent department that you will ever deal with, and they are in desperate need of anything "scientific" to make themselves look astute at what they do. What we end up with are algorithms that promise to make hiring easier, but in reality just make it even more unfair than is now.
84
I agree with the previous poster. HR is one of the most inept departments in an organization. Its ironic that the first gate to a company is staffed with incompetent people.
52
My previous company put someone completely unsuited into HR. Why? Because they could bully him into doing whatever they wanted done. HR is one of the most useless departments in a company. They have no idea what the requirements are for the jobs they are posting. They protect the company they work for not the people working for the company. They are not trustworthy. They lie to the employees. They have no understanding of how to handle difficult situations except to say that they'll look into it.
HR departments have hurt more people than they have helped. In one case, even though people complained about an employee posting disgusting pictures about abortion (he was against it) and other offensive religious items they did nothing about it. But if we came in wearing jeans during a snow storm that was a problem. I don't think they have any idea of the resentment and anger they stirred up or of how much they were not told. All of that festered into turning the company into a very unpleasant and unhappy place to work. I'm so glad I'm not there.
HR departments have hurt more people than they have helped. In one case, even though people complained about an employee posting disgusting pictures about abortion (he was against it) and other offensive religious items they did nothing about it. But if we came in wearing jeans during a snow storm that was a problem. I don't think they have any idea of the resentment and anger they stirred up or of how much they were not told. All of that festered into turning the company into a very unpleasant and unhappy place to work. I'm so glad I'm not there.
36
The article mentions an unconventional applicant who was hired at Rackspace based in part on algorithmic analysis, after being previously rejected. What it doesn't say is whether the employee was highly successful there, without which information it's hard to evaluate whether that is a positive or negative example.
51