Transcript

Important: Our transcripts at HRExaminer are AI-powered (and quite accurate) but there are still instances where the robots get confused and make errors. Please expect some inaccuracies as you read through the text of this conversation. Thank you for your understanding.

FULL TRANSCRIPT (with timecode)

00:00:09;21 – 00:13:04;10

DS: I’m David Shadovitz, editor of Human Resource Executive magazine and I’m here at the recruiting trends and talent tech conference in West Palm Beach Florida with John Spencer who is the founder and principal analyst with HR Examiner.

DS: John thank you for joining me today.

JS: Thanks for having me Dave. It’s nice to be here.

DS: Yeah I know you just finished this report on AI, how long have you been working on this report by the way,

JS: We started the research in late May and published the report right around the first of October so. So it was almost five solid months of work.

DS: okay. And I know in your summary you talked about the fact that no one in HR is really doing AI yet. Could you explain about that and also how are you defining AI.

JS: So as part of the research. I went and visited the heads of the AI departments at Berkeley, Stanford, MIT, and the University of Toronto, which are the AI centers academically. And I had a conversation with each of those guys about what AI is. And in their view AI is an intelligence that you could have a conversation with and the conversation that you have could include things like. The intelligence saying I don’t want to talk to you anymore or changing the subject or noticing something different about what you don’t like it like intelligence. The idea of artificial intelligence is that it’s intelligent, The things that people are doing right now are amazing the the using the tools that may end up with an intelligence like that 20 or 30 years from now is a great set of experiments. But when you call it a are you detract from the value that’s being generated. And so when I say that nobody’s doing anything I mean is. My count today is there one hundred and thirty companies claiming some sort of a capacity in H.R. technology. None of them have it. And when they say that they are they they distract you from what they’re actually doing that’s smart.

DS: You say you you are using the term intelligence software right as the description for what is happening in the function.

DS: Could you explain what you mean by that?

JS: Yeah. So intelligent software that we are in we are fully in the second generation of computing technology and it is profoundly different from the first generation. The epitome of the first generation was applications that wanted you to spend more and more time and we spent a lot of but we spent a lot of years working on intuitive interfaces that were sticky that you would go to and spend more time in intelligence software doesn’t want you in the software intelligence software wants you to be out doing your job. So it does things for you that are smarter than the old stuff. Would original software was like a blank spreadsheet that you filled in with data and then you recalled and evaluated the data. Intelligence software creates its own data. Because they’re looking at the recruiting function then what are some of the exciting things that you’re seeing happening there as it applies to intelligence software. So there there are four or five basic use cases being developed inside of recruiting. Chat bots that do screening. Résumé sifting and sorting tools appraisal tools workflow automation. And a basket of goods that aren’t quite clear just yet. And the exciting stuff that’s happening there is that as people start to automate decision making in low level things it allows us to imagine other things. So as the processors get automated the HR silos start to collapse and you start to see things like. Onboarding. And employment branding aren’t such different things really what you want out of a recruiting process is to have. A productive worker as soon as possible. At the end of the funnel and recruiting and if you use employment branding as the initiation of onboarding so people are being trained from the first moment they start talking with the company you end up with a shorter cycle having a productive person on.

DS: So if you’re an employer and you’re looking at leveraging these capabilities what are the key challenges and what are some of the risks that might be involved in that?

JS: So the first thing to note is that the maximum benefit of a new technology always accrues to early adopters early adopters also take the largest share of risk and so so the risks are. For instance in reservoir a extraction. The risk is that if you force your process to only consider qualifications you’ll never find Albert Einstein. Albert Einstein simply didn’t have the qualifications to do this stuff that he was great at. So when you sift to serve a resume maze and get really good at establishing a hard cut line you’ll leave serendipity out of the process. And that’s that’s probably the largest risk goodness in this area.

DS: And I know you also talk about ethical considerations that need to be sorted out. Explain what you mean by that.

JS: Well so. So there are several dimensions of ethical consideration this bias and there is what happens when you have machines making decisions. There is I think I actually think that probably the biggest thing you’re going to see in the ethical area is that HR software is going to start having product liability associated with it. To this day in time software hasn’t have liability but if you start to damage people with software and that’s what we’re headed towards they’ll be recoverable damages associated with that. And the HR department or the software company is going to have to cough up the ethical question is. So it’s it’s easy to do by analogy if I’m driving down the road and Google tells me to turn left turn left. And I don’t ever argue with Google. I get lost. I don’t know about you but Google gets me lost periodically and still I don’t argue with Google.

JS: If you’re a recruiter and the machine says here’s the top term resume it is out of the thousand we just look to look through. The odds that you’re going to go back and look through those thousand razor blades to prove that the machine is wrong or about zero which means that even though the vendors will describe that as a recommendation the machine is fundamentally making decisions about human beings and when you have machines making about decisions about human beings you’ve got an ethical problem.

DS: Now, I know in your report you talk a lot about the vendor landscape and you have startup vendors smaller vendors like that and then you have some of their more prominent larger players, who should the HR leader bet on in terms of taking this technology and implementing it in their organizations?

JS: It’s an interesting question. So, so, unlike the last several generations of technologies bigger companies have an advantage in order to do this stuff you need data. And startups have to scrounge hard to come up with adequate data to do their work with where the existing companies already have data in their war chests. So they get this great advantage just by being who they are. At the same time the likelihood that a big company is going to come up with a radically innovative construct that’s a lower percentage probability. So if you’re a buyer you have to really look carefully at both. You can’t make the assumption that you might’ve made 10 years ago that innovation is going to come exclusively from startups. It’s going to come from a broader spectrum of things.

DS: Okay, so along those same lines as HR executives, as employers begin to evaluate evaluate this software. What are some of the key questions they need to be asking these these vendors?

JS: Well the first thing is if they’re telling you that it’AI it isn’t and you probably need to just deal with that with them upfront. The question that you want to know the answer to is. What does this stuff do. So that’s the first, what’s the point of this what do I get out of. What’s the return. The same standard questions just then It’s very difficult for an employee to overrule a decision made by a machine. And so it’s really important to understand what the methods are required for arguing with the decision that the machine produces. And How do you do that. The biggest problem with these tools is what I refer to as latency and that’s how long does it take the machine to learn a new thing. Organizations change dramatically in unpredictable ways. With things like CEO transitions and stock price volatility and competition and workforce changes. And every time those changes happen the way that decisions get made inside of the organizations vary.

JS: Machines are designed to make decisions in a repeatable way and so the point of machine learning is to get the machine up to the current standard. But what I just said is that standard changes all the time. So what you want to know is how long does it take the machine to recover from a change, because in the time that the change is happening, the quality of the decisions coming out of the machine is going to be lower than it is when it’s back up and all the way learned. So you want the you want the vendor to explain how that works in your environment.

DS: And I mean it sounds like you know the rules are really changing in terms of the kinds of questions in light of this new technology that the HR leaders need to be prepared to be asking the vendors and assessing the products through those questions.

JS: So we’re in the early days and the questions that are interesting today probably are going to be interesting three or four years from now. It’s my sense if you look at the big players Oracle SAPm Workday, ultimate cornerstone, ADP that kind of player, they are all developing Personalities in their intelligent systems. And it’s my sense that what you’re going to do five years from now is more like dating the software then it is like evaluating its functionality.

DS: And I guess one final question you know based on the research that you did what was this single thing that surprised you the most going into it you expected something maybe totally different.

JS: I really I thought I was going to find artificial intelligence. I thought I thought I was but the point of the research was to evaluate artificial intelligence and HR and that’s an interesting shorthand for what’s going on but it’s not what’s going on so that so, the very premise of the of the report was wrong.

DS: Interesting. Well we’ll end there. Thank you so much John for joining me today.

JS: Thanks so much David, this was great.