Transcript
Mark:
Welcome to PeopleTech, the podcast of the HCM Technology Report. I’m Mark Feffer.
My guest today is Gershon Goren, the CEO of Cangrade. They create models for success, interview tools, and onboarding and upskilling solutions. We’re going to talk about data in HR, the evolution of AI, and where it’s all headed, on this edition of PeopleTech.
Hi, Gershon. Thanks for coming in. Can you tell me about Cangrade? What do you folks do?
Gershon:
Our mission in the world is to create data powered recruitment and talent management. We started in the world where AI was not yet a thing, it was in 2014. Or it was a thing, but was not mentioned a lot. We initially labeled it as really predictive analytics, but what we do is using machine learning techniques in order to accurately and bias-free predict the possible success of candidates in jobs.
Mark:
Now, a lot about AI in talent acquisition has been in the news lately, especially with chat AI’s efforts really going like wildfire. Excuse me. But the attention to AI seems to happen every few years. And I wonder, what’s your read of it all this time around? Is this a true wave of true AI or is it just marketing speak?
Gershon:
I think it’s a little bit of both. And clearly what we’re seeing lately with this large language models is a big leap into the new era. And I think we should totally be ready and expect for AI to become a significant part of our life. And it will impact us in all kinds of way which are impossible to predict, including what will happen. In field of talent management, it’s definitely been long coming, and it only might seem like something new to people who were not paying attention to this field before, which is totally fine. It was definitely something that was very much used by companies for at least decade in different versions. But large language model type of AI, the generative AI definitely made a lot of headlines. Frankly, because it seems more capable and almost sentient, which it’s not. But it’s ability to have basically a response that resembles human responses is definitely very intriguing and has definitely a lot of implications going forward.
Mark:
How much of this is about changes on the customer side, on the user side? The developers of AI applications are always going to be out there pushing their products and their work, but is the audience more receptive this time?
Gershon:
Clearly, clearly. Just judging from ChatGPT traction on the market, and this immense user base that they were able to acquire in a very short period of time, this was record-breaking. People are clearly paying attention to this. People are clearly starting to adopt it in all kinds of ways. And I think it’s only the beginning, truly, for this type of technology to, again, impact us in all kinds of ways. This is only early days of it.
Mark:
We also have big brand names involved this time more openly, and I’m thinking about Microsoft and Google. And I know there are others. What impact has that had? How does that fit into the equation?
Gershon:
You mean in talent management?
Mark:
Yeah.
Gershon:
I don’t think anyone has a clear picture yet how it impacts talent management. And what is very clear to me that in talent management the adoption will not be super fast. In big part, because of the recognition of the potential risks attached to this type of technology, possible biases and inaccuracies. And I think it’s well merited concerns, and I expect this area to be regulated. And there’s going to be a lot of compliance questions with regards to this type of technology. I think actually HR will move to this space with somewhat more caution than other fields, but there are parts of HR that will probably adopt it faster than others. And I think, actually, the sourcing will be very impacted by this in very short-term, because sourcing of candidates is in many ways similar to marketing. And marketing is definitely getting supercharged by the ability to personalize content and personalize messaging to the audience.
And that’s exactly what will happen for essentially sourcing candidates, and that’s probably a good thing. I think there’s going to be slower progress when it comes to talent selection and screening. There’s definitely going to be more risks in using this type of technology for identifying the right fit, because again, all the possible ways in which it can go wrong. The area where I think it’s going to be really exciting is actually development and learning and onboarding, that’s where possibilities are just mind-blowing. And I think the slight little longer term, what will happen is there’s definitely going to be a use of AI for selecting candidates and sourcing candidates, and the criteria will probably be broadened for the type of talents that will consider it to be qualified. The criteria will be different.
It’s not necessarily going to be based on resume or even experience, but it will be more potential driven. And the training and onboarding and ramping up the hires will start immediately. And it will be just much more personalized and much more, I guess, faster probably from the moment you accept the offer. It will start, and it will be primarily driven by this AI technology. The expense from company’s perspective on training the new employees will be vastly diminished.
Mark:
What about bias? That’s a common concern today in AI, but how does that fit in or what kind of concern is that given that-
Gershon:
I think it’s a huge concern. And I think biases is something that is very endemic to AI technologies in general, and our differences between different types of AI. And in some types of AI, it’s easier to control for biases than others. Specifically, I want to delineate between what is known as narrow AI, which is actually something that has been already in use by many companies for at least a decade. And that is, for the purpose of this conversation actually, what Cangrade does. Narrow AI focuses on very specific types of data, and very specific types of outputs that it’s aiming to predict in this types of situations when data is more well-defined, it’s also easier to control for biases and essentially drive them out. In generative AI, which is completely generic. And it learns on, for the lack of better terms, on everything that exists on the internet, it’s order of magnitude harder to predict all the ways in which this can go wrong and biases can creep in.
For that purpose, actually, I don’t envision generative AI to be particularly useful in questions that have to do with essentially decisions about candidates or employees, about their placement, about their hiring. I think it’s still a long shot before this type of technology can actually reliably make these types of decisions and not risking some really spectacular failures and biases. And this is already something that happened. That it was attempted, it was already deemed as potentially dangerous. I’m sure you’ve heard about Amazon doing something along those lines a few years ago, and only to realize that there were severely discriminating female employees in doing so.
I think actually in the talent acquisition, candidate selection part, the narrow AI will still dominate the field. It’s much more reliable, it’s much more explainable. The data types are much more specific, and they can be easily demonstrated as something that is valid and indeed predicting employee success. That area is better understood, and ways in which biases can be reigned in are also existing. And in fact, Cangrade even higher filed a patent for cleaning biases from this type of selection algorithms, and obviously others doing similar things. That is what my money is on, for at least short-term. But of course, this whole area is evolving so fast that it’s, as I already said, hard to predict what the generative AI will be capable of even a year from now.
Mark:
I want to shift gears just a little bit and talk about the users, the recruiters. Are they ready for this? Do they understand the technology well enough? Or will it be hidden from them, so it doesn’t matter?
Gershon:
First of all, it matters. It matters. And I disagree that users need to see this as a useful black box and not to concern themselves with how it works. I think every organization, at least on the organizational level, needs to have the expertise in order to understand what this technology is doing. Otherwise, they will be sold something that is potentially not useful and dangerous. It’s something that every organization should be actually vetting with the level of knowledge that makes them comfortable that they’re buying the right products and using them for the right purposes.
Is this how it is today in the industry? No, clearly it’s not the case. Obviously we’re working with customers, and we’re seeing different level of expertise in this area. Unfortunately, more on the side of lacking the expertise. And clearly, HR is not ready for this. HR is not known for being particularly advanced part of organizations, for most parts. There are some notable exceptions. And that will need to change. But that said, I think nobody is ready for this technology, nobody is truly prepared. And I think that’s a problem on the societal level for us, not just in HR.
Mark:
With that said, can you sketch out the next, say 36 to 48 months for me, how you see this all proceeding and expanding over that time?
Gershon:
I think there’s going to be a lot of attempts to AI generative, and there’s going to be definitely a lot of attention on generative AI because this is all the rage right now. And I think some of this attempts will be unsuccessful. I think there’s going to be a lot of noise on the market, a lot of new offering just popping up overnight. Because again, everything is moving so fast. And frankly, I think there’s going to be hard for companies to make sense of everything that all of a sudden became available on the market, and truly decide what they should be adopting and what they should be staying away from. Because there’s definitely going to be a range of solutions, not all of them are going to be good. In fact, I think minority of them will be in fact good and providing value for companies.
There’s also a clear distinction between something that is purely cost saving. And for example, you can use something like a resume scanner, whatever you call it, something that analyzes the resumes. And deciding, based on the resume, whether somebody’s a good fit or not. I personally think it’s a type of technology that can be definitely cost saving, and allowing recruiters to not do it manually. But is it really a good technology that helps organizations to achieve better results in hiring? I believe not. I think this, again, a lot of potential biases involved in this process. And one additional factor that we all need to keep in mind that AI is now available not only for recruiters, first and foremost, it’s available for job seekers. They will use it, they will use it intensively. And this is going to become an arms race of sorts, where basically candidates will optimize their CVs to better fit the job descriptions. And companies will try to optimize their selection based on this resumes to, again, find somebody who feeds better to their job selection.
I think the results are going to be less than ideal. And this is just one example of such what I consider to be misuse of AI, or not productive use of AI. And I think it’s going to be a daunting job for companies to actually understand what is just cost saving but not beneficial for the sake of company’s success in hiring. And what is actually something that potentially not as cost saving, but will impact their bottom line in terms of hiring best fit candidates, or what is potentially both cost saving and beneficial for the end result.
Yeah, it’s going to be challenging. It’s going to be challenging couple years, for sure. Because as you said, there are some big companies who are jumping into the fray, for sure. Talent management is not their main attention, and it will not be. Because again, I think this is the area where it’s not so clear what is the right solution right now. Maybe for some exceptions. As I mentioned, I think in the talent development it’s definitely going to be pretty clear the kind of stuff that will need to happen. In talent sourcing, a little less clear, like in many marketing adjunct space. But in talent selection, it’s definitely going to be potentially dangerous. The attempts will be made in all of this areas above, and I think it’s going to be every company’s job to see through it and to understand what is the right solution for them.
Mark:
Well, Gershon, thanks very much for spending your time with me this morning. It was great to talk with you. Great to meet you.
Gershon:
Yeah, same here.
Mark:
My guest today has been Gershon Goren, the CEO of Cangrade. And this has been PeopleTech, the podcast of the HCM Technology Report. We’re a publication of Recruiting Daily. We’re also a part of Evergreen Podcasts. To see all of their programs, visit www.evergreenpodcasts.com. And to keep up with HR technology, visit the HCM Technology Report every day. We’re the most trusted source of news in the HR tech industry. Find us at www.hcmtechnologyreport.com. I’m Mark Feffer.
Image: iStock
