AI: What Every HR Dept Needs to Know
Podcast & Transcript: Uncertainty can cause an employer to kind of "freeze in place" and be reluctant to advance the use of AI technology
cybersecurity, payroll and other employment areas. This podcast takes an in-depth look at AI issues with Garry Mathiason,
who spearheaded a first-of-its-kind practice area in forming the Robotics Practice Group at Littler Mendelson.
PODCAST & TRANSCRIPT (excerpts)
Weisenfeld: There’s no question artificial intelligence is poised to change the workplace at a level not seen since the Industrial Revolution, and is already doing so in many cases. But what do these changes mean on the job front and how are they affecting the hiring process?
On this podcast we’ll take an in-depth look at AI with a man at the forefront of this trending area, Garry Mathiason. Garry spearheaded a first-of-its-kind practice area in forming the Robotics Practice Group at Littler Mendelson, the nation’s largest employment law firm, to address the many compliance issues arising with these technological advances.
Weisenfeld: Well Garry, so much is being written about AI these days, as we were just talking about off the air. How are companies using it on the employment front?
Mathiason: David, there’s a tremendous expansion in the way AI is currently being used, and in the near future will be more intensively used, in dealing with companies on the HR and employment front. Obviously you’ve got a tremendous number of factors going on in the recruiting process. Aspects of that that are expanding is the ability to reach into potential candidates that aren’t necessarily on the job market but are likely to be interested in potential job offers.
There’s also just the ability to work through hundreds of thousands, or sometimes millions, of applications that are on websites, something that you couldn’t do physically but can do with AI.
Then on a more sophisticated level, AI is actually breaking down the job requirements and looking at résumés and breaking those down into their elements, and doing a much better job of connecting the necessary skills with the available positions.
That’s just in recruiting. You have then cybersecurity, you have AI working in the back office of HR from payroll to a number of routine functions that are especially well-suited for that technology. It’s advancing in the area of training. It’s actually entering the field of interviewing, and I’ll be talking a little bit about video interviewing and how AI relates to that. It’s reaching into the performance areas during the work day.
And it’s also looking at the potential that an employee is considering resigning or likely to take another position. AI can give you an indication of how likely that is. You also have some surprise analysis where you’ll find people that are kind of hidden within the structure of an organization but are excellent performers, and AI can sort them out.
This is just the tip of the iceberg. I think the entire HR function is being reviewed from this perspective of AI, and that in no way pushes against the value and importance of human beings, especially in HR. But it really augments, facilitates and expedites, and I think in many cases improves the processes that we deal with.
Weisenfeld: I definitely want to get to some of those recruiting and hiring points that you mentioned, but before I do, I was interested to hear you say recently that HR has been slower to adopt AI tools than the rest of organizations, including even at the C-suite level, and that they can sometimes be mired in the past. So tell us a little bit about what you meant by that and what HR can do better.
Mathiason: Well I think if you look at the C-suite there’s hardly a major company on the horizon that doesn’t have an AI plan. And even in the C-suite there’s a lot of consideration about how technology, advanced technology, is going to transform whatever process or product or system that the company is working with.
Back in 2017 there was a survey done that showed that 52% of HR professionals looking at their company didn’t think there was going to be the use of AI and weren’t anticipating it, which is really stunning because at the same time in the C-suite that percentage was probably down to less than 25%.
But to give you an indication of the change, we estimate that about 10-20% of HR organizations within companies are approaching something called the ‘mature use of the technology.’ That leaves a lot of room to develop. In sales that number is over 50%.
I do predict in the next five years we will make up the gap in HR because there’s such an awareness going on right now of how AI can be used and there’s a demand from the C-suite that this occur and occur rapidly, and I see the transformation coming day by day.
Weisenfeld: Well Garry, with recruiting and hiring, robots have been able to do some of these functions certainly for a while, but can they simulate a human interviewer’s ability to read candidates and their cues, such as considering the benefits, for instance, of a diverse workforce? [0:06:13.1]
Mathiason: Remarkably, AI working with human interviewers can be even more effective than what the human can do themselves. There’s a very, very interesting development, very recent development, where there’s a lot of technology coming out that will take a video interview and AI will look at a variety of characteristics that occur during the video interview – not just the content of the words expressed but the facial expressions, the pauses, the gestures – and will actually give you an assessment of how they view the candidate.
There are some legal considerations to be looked at but this is maturing, and is increasingly being recognized as something that is a valuable tool. And with that you can actually help expand the diverse workforce by taking away with the technology some of the indicators that might cause unintentional or unconscious bias on the part of a human interviewer, and I think that’s a plus.
But all of it needs to be balanced because there still are many factors that the human being brings into bear that AI can’t simulate. So it’s a combination of the two.
What I urge, however, is to be very careful to look at making sure under existing laws, that were written really before the technology was developed, whether you are reasonably in compliance. And that’s one of the initiatives we’re undertaking is to really give compliance solutions to companies so that they can move forward and use the technology more effectively and more extensively to accomplish the goals of the organization.
Weisenfeld: You mentioned unconscious bias and speaking of that, some say the technology here is only as neutral as the humans who are putting the algorithms together and that there are hidden biases that can plague AI selection tools as well. What are your thoughts on those concerns?
Mathiason: It is a two-way street. There are several areas in which this technology can actually help reduce bias but there are definitely areas where the bias appears. If you go to Wikipedia there’s about 178 different forms of bias associated with AI. There’s no question that the original programmers, the people putting in the code, bring with them a certain characteristic or way of looking at life that sometimes gets reflected into the algorithm. So there is the potential of bias there.
Probably of more concern is with machine learning and the fact that increasingly the technology will look for patterns and compare patterns on a basis that we’ve never before been able to do. That might be a million different factors all considered very quickly by AI. Sometimes what appear to be benign characteristics, like looking at the zip codes of successful candidates and comparing them to zip codes of applicants, can have an effect of creating a form of unintentional bias, disparate treatment, where you find that all of a sudden minority areas are less considered and the effect of it is to effectively be a form of discrimination.
There are a variety of solutions for this and you have to be conscious of them, including the fact of being able to do a review of the candidates that come through your process to find out whether that disparate impact is actually taking place, and there are data analysis systems that will actually give you that answer. And we recommend that that type of review procedure or checking procedure be used by companies that are relying on this technology. It’ll improve the outcome.
Weisenfeld: Garry, as someone who’s so immersed in this area, what’s the biggest regulatory concern – and perhaps it’s biometrics or perhaps it’s something else – that you’re hearing from clients when it comes to AI?
Mathiason: I think it is uncertainty. There is concern that the regulations will harm innovation. But what we find much more of concern is uncertainty, not knowing what the regulation is going to be, and that causes the employer to kind of freeze in place and be reluctant to advance the use of the technology. That’s one of the missions that we undertake here in the law firm with extreme vigor is to try and think through reasonable compliance systems to give our clients some better assurance that what they’re doing has a much higher probability of being compliant.
Weisenfeld: So in light of that uncertainty you just mentioned, in our final couple of minutes do you have any practical recommendations for how employers can handle that and get past it?
Mathiason: I think it is critical to do certain basic things. One is to identify the technology that you’re using. I think as an important practical matter, cybersecurity is so vital. I think the employer has to take extra steps to protect data that it would collect.
I think looking at areas where it can notify the employee of the fact that the technology is being used, that’s very important. But again my primary recommendation is to set up a method of checking to see after the fact that the technology is being used correctly and effectively. And that could be an audit of the data. It could be a number of different review systems. But that’s going to help make sure it’s right.
And one of the things that we’re encouraging is consideration of a safe harbor provision in future regulations and legislation that would indicate that employers that take some of these preventive steps would effectively be given a greater immunity towards especially private prosecution, and avoid some unnecessary lawsuits. And it’s a great trade-off because then in return for that you get self-compliance and effectively a much better system right from the beginning.
Weisenfeld: Certainly a lot to mull and to watch in the coming years, that is for sure. Garry Mathiason is the co-chair of Littler Mendelson’s Robotics, AI and Automation Practice Group. He also has authored a workplace checklist on XpertHR.com to help employers prepare for artificial intelligence issues.
FREE DOWNLOAD: Customizable Checklist:
Prepare for Workplace Robotics and Artificial Intelligence Checklist
Weisenfeld: Garry, thanks so much for joining us.
Mathiason: Thank you. It’s been enjoyable to be here and there’s a lot of excitement coming off the horizon.
Complete transcript: How Artificial Intelligence Is Revolutionizing the Workplace
See also: New Tech, New Work…New You!
Dealing with New Technologies in the Workplace