Will AI make better hiring decisions than humans?

Jun 28, 2017
George Anderson

Unilever is leaving the decision on what candidates to interview for jobs leading to management roles within the company up to artificial intelligence (AI) technology, according to a Wall Street Journal report.

The use of AI means that the consumer products giant is doing away with campus visits and resumes submitted by prospective employees. The new process makes use of algorithms to identify workers that fit Unilever’s criteria for hires. It also allows the company to expand its search beyond the small number of schools where it used to send recruiters.

Unilever uses targeted ads on Facebook and career websites to find candidates. By clicking on the ads, interested individuals are sent to an application form where Unilever pulls information from LinkedIn to populate required fields. Algorithms eliminate more than half of those applying for jobs.

Should candidates pass the initial cut, they are asked to play a group of short online games as part of a skills assessment. The games test candidates’ ability to perform under pressure and assess other skills. Those grading out in the top third submit video interviews to answer questions about situations they may face on the job. This assessment further winnows the field. Those remaining getting an interview with a human being.

The Journal reports that Unilever credits the program with better hiring outcomes. The process moves faster than when it was completely human-driven, and 80 percent of those who make it to the end receive job offers.

Unilever is not the only company turning to AI, and tech startups are popping up to help employers fill jobs.  Josh Bersin, principal at Bersin by Deloitte, told Fortune, “I get emails every day from someone who decides they’re going to fix the recruiting market through artificial intelligence.”

The use of AI is also eliminating barriers, such as the lack of a college degree, that used to keep some from finding jobs. Google, for example, has been public about the increasing number of people it hires for skilled positions even though they never went to college.

DISCSUSSION QUESTIONS: How do you see artificial intelligence affecting the hiring practices of retailers and consumer packaged goods companies in the years ahead? Do you think using AI will result in better hiring outcomes for employers and employees?

"I think AI can get you part of the way there but, boy, it’d be a huge mistake to rely solely on it. "
"I don’t think any company would forgo the in-person visit. At the very least you’d want to make sure you weren’t hiring a robot."
"A huge part of a company’s success is how well the team works together and that’s not something AI can detect."

Join the Discussion!

16 Comments on "Will AI make better hiring decisions than humans?"

Notify of

Sort by:   newest | oldest | most voted
Lee Peterson

I think AI can get you part of the way there but, boy, it’d be a huge mistake to rely solely on it. There are so many human nuances involved in a discussion, I certainly hope those would be evaluated as well. How does a machine perceive friendliness? Or concern, or kindness or reliability?? What is it all coming to — Blade Runner?

Art Suriano

We are seeing AI taking over more and more human tasks. Technology is faster, more accurate and less expensive than people. It’s understandable that AI is finding its way in the recruiting and hiring process. However it’s still new. I believe we will see modifications added and I believe some of those modifications will include more human involvement as well. Technology is a tool and, although it provides many great benefits, it still lacks certain human qualities that are important in the hiring process.

Doug Garnett

There is a fundamental problem with programs like this. Namely, there is no way to know what the program misses. Algorithms work well when they can be tested, checked and improved. But a hiring algorithm has no checks — there is no way to know if it actively rejects potential hires who are good. And that makes algorithms very dangerous to company and societal health. (This error is extensively discussed in Cathy O’Neil’s book “Weapons of Math Destruction.”)

That said, it’s not surprising that Unilever tells us how great the program is — they seem to be loved by bureaucracies for simplifying bureaucratic life. Unfortunately, Unilever will never know whether the program is effective.

I highly recommend reading O’Neil’s book in order to learn where algorithmic data applications can be effective and where they mislead.

Dr. Stephen Needel

The ability to predict career success has always been a holy grail of HR. It’s not so much that it’s AI as it is a good predictive model of career success — this could be created without the use of AI. That said, there is still some subjectivity to it. Employees need to be evaluated as input into the model. Any model needs a range of very bad to very good employees in the initial learning phase. That evaluation is likely to be somewhat subjective when we’re talking about these types of people. The risk is homogeneity — you always get the same type of people passing the screening.

Mark Ryski

As any hiring manager knows, it’s very difficult to pick winners. Traditional approaches relying on resumes, reference checks, interviews and aptitude tests are highly subjective and imprecise. Using AI as way to screen potential candidates is a very interesting approach that has the potential to improve outcomes. However, it’s not perfect either. One problem is that this approach relies heavily on social media as an input and while many people do use social media, those who don’t will be excluded from the process. AI is being applied in all aspects of business and recruitment is just another and important one. Ultimately, I believe that it will produce better outcomes for both employers and employees.

Ralph Jacobson

As far back as the late 1980s, I can remember using rudimentary tools for predicting potential performance and tenure of prospective employees at my stores that I managed. As technology has evolved, AI at its core is truly augmenting human decision making with the capability to consume vast amounts of data far beyond any human ability, and offer increasingly accurate suggestions as more data is consumed. This is absolutely a great way to leverage this machine learning technology.

Joan Treistman

In the past there have been services that ask a prospective employee to complete a paper survey populated with questions to determine if the prospect would be a good fit. Importantly, the job criteria were identified and the corporate culture addressed to some extent. The analysis, provided at a cost of course, would identify the pros and cons of hiring that individual. And all along there have been psychologists and sociologists who would review candidates for a corporate position.

So that was then. Now it’s online with all the efficiencies and comprehensiveness the selection process requires. I don’t think any company would forgo the in-person visit. At the very least you’d want to make sure you weren’t hiring a robot.

Cristian Grossmann

AI may be a good tool for the first round to capture keywords that meet your minimum requirements, but I don’t think it should go further than that. A huge part of a company’s success is how well the team works together and that’s not something AI can detect. It still lacks emotional intelligence compared to humans. And if you’re hiring for a customer-facing role, AI especially shouldn’t be solely relied upon. There are so many instances of candidates not being the exact fit on paper, but because of their work ethic and attitude they will excel must faster than someone who met the technical skills. And that’s something humans can pick up on in interviews.

Lee Kent

Nowadays, especially as related to delivering the right customer experience, culture is king. While AI can possibly help make the first cut, unless you are looking for robots AI can’t assess a culture fit. Not yet, anyway!

And that’s my 2 cents.

Mel Kleiman

Two quick comments:

1. AI is wonderful at predicting things, but people are different in many ways and they don’t come with manuals or directions on how to get them to be more productive.

2. The real difference between average and great performers is that great performers do the things they do not like or want to do that need to be done to be great. If we can get AI to tell us who will not only have the skill and ability, but more importantly thrive in specific environments that are always changing, then maybe that will be a real game changer.

gordon arnold

This discussion identifies one of the most incredible company investments of the early 21st century. By all accounts that I have read and visited, the money being spent here is growing at a phenomenal pace.

Corporations of all kinds wishing to minimize overhead costs look at human resources as a place to automate for saving time, increasing productivity and improving the work place. The interesting obvious testimony to the success or lack of comes from the very industry from where it is built and supported. Information Technology companies are seeing the same results that retailers are and in many cases, the results are even worse. That says a lot about finding, placing and keeping the right people with Artificial Intelligence.

My candid observations have seen low performance candidates from inside and outside the company given opportunities that were far ahead of their abilities. These opportunities were supported simply by verbal and/or written override to the system.

I think we are still buying the sizzle with no sign of the steak.

Craig Sundstrom

I’m unclear on what’s changing here. HR departments have always used various tests and strategies for selecting people, so this just seems like the latest “new thing.” Mention “AI” and suddenly everyone is excited over the old wine in new bottles.

As for whether or not companies will be “better” at hiring employees, I don’t think that’s the problem, most of the discussions on RW center on problems companies have with making effective use of (good) people after they’ve hired them, not before.

Ricardo Belmar

I think this is yet another great example where AI is eliminating the need for humans to handle the more time consuming, yet low reward tasks. In this case, it’s helping HR eliminate early candidates based on rules they most likely had people following to produce the same number of eliminations, but in a fraction of the time. This will surely reduce hiring and recruiting costs. I don’t see how AI would replace the final human decisions about who to hire and who not to hire as this is just to complex and a decision that often ends up with someone following a “feeling” or a “hunch” about how the right choice should be. I think we are a long way from AI being able to handle that.

Shep Hyken

AI can be very helpful in the application process. I like using behavioral style assessments as part of the application process, and there are plenty out there. They can be very helpful in determining the potential good fit of an employee. It has always been up to a person to look at the results and make decisions. AI can help eliminate that step. It can administer the assessment and deliver a report. It can compare and contrast styles based on past assessments of successful candidates — probably better than a human.

However, after that part of the process, there still needs to be an interview. More than just mindset, there is personality and attitude (and probably a few other considerations). Those need to be in alignment with the culture of the organization. AI can support and help, but at this time, shouldn’t be the final decision maker.

Vahe Katros

“Software Is Eating the World, But AI Is Going To Eat Software.”

Jensen Huang CEO, Nvidia

William Hogben

A common mistake with hiring is to let the interviewer make the final decision. This causes all sorts of mistakes and results in demonstrably worse outcomes, according to studies of college admissions processes, something not dissimilar from hiring decisions.

The solution is to have the interviewer score the candidate on several standard criteria, then pass their scores and notes on to a second person who’s never met the candidate, and makes the decision incorporating the interviewer’s notes as only one factor. This role can be handled by AI similarly well.

"I think AI can get you part of the way there but, boy, it’d be a huge mistake to rely solely on it. "
"I don’t think any company would forgo the in-person visit. At the very least you’d want to make sure you weren’t hiring a robot."
"A huge part of a company’s success is how well the team works together and that’s not something AI can detect."

Take Our Instant Poll

How effective is artificial intelligence as a tool for finding the best candidates for available jobs?

View Results

Loading ... Loading ...