Ep 24: Driving Better Hiring Outcomes through Talent Intelligence and Responsible AI with Markellos Diorinos

[00:00:00] [SPEAKER_00]: You can contact us via a lot of people who are going to be a nice day to join you.

[00:00:06] [SPEAKER_00]: But with Sendesk, we're going to do a better customer experience.

[00:00:09] [SPEAKER_00]: Better for the big mother, better for the family, better for the young man in house number 3, better for you, better for all.

[00:00:17] [SPEAKER_00]: Because we have a few hope that the customer has always been right, we're going to be the customer in a moment.

[00:00:22] [SPEAKER_00]: And we're also going to be the best customer for everything.

[00:00:26] [SPEAKER_00]: Sendesk, customer experience with KI is always loved.

[00:00:29] [SPEAKER_01]: Feeling kind of left out at work on Monday morning?

[00:00:33] [SPEAKER_01]: Check out the barf, breaking news, acquisitions, research and funding.

[00:00:37] [SPEAKER_01]: It's a look back at the week that was so you can prepare for the week that is.

[00:00:43] [SPEAKER_01]: Subscribe on your favorite podcast app.

[00:00:55] [SPEAKER_04]: Hey, it's Bob Pover. In this episode I'm joined by Markellos Diorinos, co-founder and CEO of Rick,

[00:01:01] [SPEAKER_04]: the hiring intelligence and talent assessment platform.

[00:01:04] [SPEAKER_04]: We dive into the world of data driven decision making and hiring,

[00:01:07] [SPEAKER_04]: exploring how talent intelligence can revolutionize the way companies match individualist roles.

[00:01:12] [SPEAKER_04]: Our fellow shares insights on mitigating AI biases,

[00:01:15] [SPEAKER_04]: the importance of third party audits, and the challenges of regulating AI in different regions.

[00:01:20] [SPEAKER_04]: Our calluses, they fellow responsible AI advocate and joins me from its home in Athens, Greece.

[00:01:25] [SPEAKER_04]: So it is a great perspective on this space. We also discuss the concept of AIQ

[00:01:28] [SPEAKER_04]: and now it's becoming increasingly important in today's AI driven workplaces.

[00:01:32] [SPEAKER_04]: Hope you enjoyed this discussion. Thanks for tuning in.

[00:01:37] [SPEAKER_04]: Hello everyone. Welcome to another episode of LVH or AIQ. I'm your host Bob Pover.

[00:01:42] [SPEAKER_04]: With me today is Markellos Diorinos.

[00:01:46] [SPEAKER_04]: Hey, doing Markellos?

[00:01:47] [SPEAKER_04]: I'm doing great. Well, thank you for having me. Thank you for being here and appreciate your time.

[00:01:54] [SPEAKER_04]: Markellos, why don't you give our listeners a quick little bio about yourself and your background?

[00:02:01] [SPEAKER_02]: I'm going to be very happy to. I started my formal education,

[00:02:04] [SPEAKER_02]: I think that's where people kind of start to formulate who they are and what they want to do in life.

[00:02:09] [SPEAKER_02]: As a computer scientist, a long, long time ago,

[00:02:13] [SPEAKER_02]: that was let's put it eloquently in the previous relay.

[00:02:19] [SPEAKER_02]: And I enjoy that very much because I like the solve problem.

[00:02:23] [SPEAKER_02]: But you know as I started working as a computer scientist, developers of her acting model stuff,

[00:02:28] [SPEAKER_02]: I came to the harsh realization that hard part about software projects wasn't actually

[00:02:35] [SPEAKER_02]: called in it or solving it. It was getting people to use things.

[00:02:40] [SPEAKER_02]: And that was very, very interesting. So over time that actually made into drift away and go over

[00:02:47] [SPEAKER_02]: to the dark side, to give a fair order of business side as we like to see more of a local.

[00:02:52] [SPEAKER_02]: I thought you had the backwards.

[00:02:58] [SPEAKER_02]: And you know, I spent a bunch of years working for places like Microsoft.

[00:03:03] [SPEAKER_02]: I had the pleasure of doing product management for internet explorer, way back in the day of I7.

[00:03:09] [SPEAKER_02]: And nowadays I don't know if a lot of people know what I used to be anymore.

[00:03:14] [SPEAKER_02]: So telling them what I said and they're like, what was that? Was it like your spectrum or something?

[00:03:20] [SPEAKER_02]: I know our kids would not have no idea what that is.

[00:03:24] [SPEAKER_02]: And frankly, having been the product months of render and explorer server, I loved it to death.

[00:03:29] [SPEAKER_02]: It was excellent at the time. Nowadays it's end of time that people don't know about it.

[00:03:34] [SPEAKER_02]: They're bigger and better things for people to work with.

[00:03:39] [SPEAKER_02]: But as I was doing that, Microsoft started getting into the concept of big data.

[00:03:44] [SPEAKER_02]: And then my next occupation at the beginning has been over 10 years working with some telecoparators.

[00:03:51] [SPEAKER_02]: And everything we did at telecosuores on Masi, it was like okay, I have here 50 million subscribers

[00:03:57] [SPEAKER_02]: and they need to give them better experience or sell them up, sell them, process them, whatever.

[00:04:03] [SPEAKER_02]: And that really manifested from the mindset of big data.

[00:04:09] [SPEAKER_02]: What I learned during the time working there was that

[00:04:12] [SPEAKER_02]: I have a lot of personal bias and I'm usually wrong in what I mean by that.

[00:04:17] [SPEAKER_02]: It's not that I'm wrong over time. But if I'm looking at the solution and I'm judging it and saying,

[00:04:23] [SPEAKER_02]: yeah, you know, we have three bias, this is probably going to be the one that wins.

[00:04:27] [SPEAKER_02]: A lot of times my personal bias doesn't translate well to another 15 million people.

[00:04:31] [SPEAKER_02]: But the data always tells me that truth. So there was an lightning and the bit of sobering experience,

[00:04:38] [SPEAKER_02]: realized that you know, you're thinking a lot of things and then you realize that too.

[00:04:42] [SPEAKER_02]: What I actually know is how to ask the right questions and interpret the data.

[00:04:47] [SPEAKER_02]: I shouldn't be making so many gut decisions because I'm bias myself and I'm not always right.

[00:04:54] [SPEAKER_02]: And that's kind of, that was the first step into what later became great because with two of my

[00:04:59] [SPEAKER_02]: co-founders, one of them has a psych background and the other one has a tech background,

[00:05:03] [SPEAKER_02]: and has a real tech background, he actually does tech stuff, not like where I'm the fake data

[00:05:08] [SPEAKER_02]: nowadays. And we were hiring people and not every hire was a success.

[00:05:16] [SPEAKER_02]: And one of these days we're sitting together and we're doing a radio spec on a specific person. We're

[00:05:20] [SPEAKER_02]: saying everything we asked in the interview will be gutter, right answers, everything looked so great,

[00:05:25] [SPEAKER_02]: everything looked so good. How could we have prevented that hire? How could we have recognized

[00:05:30] [SPEAKER_02]: very good that this person was both wrong for their old hands, really really wrong for the company

[00:05:36] [SPEAKER_02]: DNA. So here we didn't, you know, didn't align with what they wanted to do. And one of the

[00:05:43] [SPEAKER_02]: co-founders generally said, look, there's a whole body of science that does that. Let's go and

[00:05:49] [SPEAKER_02]: see what two Iosecology tools are out there. And we're like, oh yeah, we can, we'll never make

[00:05:54] [SPEAKER_02]: wrong hire again. And when they looked at the tools and what we realized was that there's a whole

[00:06:01] [SPEAKER_02]: great body of science, but that's about how we should select people and what's important,

[00:06:05] [SPEAKER_02]: what to do and what not to do. That gives out a ton of data. And then you need a small

[00:06:11] [SPEAKER_02]: arm if people can interpret the data because it comes very low and it tells you this person's

[00:06:16] [SPEAKER_02]: like they seem like that, but really should I hire? Should I not hire? What do I need to

[00:06:23] [SPEAKER_02]: this one questions that were in answer and I'm also, by the way, a lot of the experience when

[00:06:29] [SPEAKER_02]: you're doing those second metrics. It literally felt like punching cards in the computer.

[00:06:34] [SPEAKER_02]: It was so antiquated, the user interface, you know, the assessments that we saw there.

[00:06:39] [SPEAKER_02]: And that's where we realized that there is a huge opportunity in finding second metrics,

[00:06:46] [SPEAKER_02]: getting big data about what the company cares about. So not having second metrics in the

[00:06:51] [SPEAKER_02]: don't exist for themselves, but how do these second metrics or other data that we have? By the

[00:06:56] [SPEAKER_02]: ways, like, medical just that the circle point cards skills, you know, whatever have you. How

[00:07:01] [SPEAKER_02]: did they relate to business outcomes? Is it performance, is tenure, is going to be a really

[00:07:07] [SPEAKER_02]: learning thing, whatever it is that I need. And we said that a creative platform, this is what

[00:07:12] [SPEAKER_02]: we did with break that actually masters that it ingest all the data from people, it ingest all the

[00:07:18] [SPEAKER_02]: data data about performance or other business outcomes. And it comes back in office,

[00:07:24] [SPEAKER_02]: practical solution that says hey, here's what works the day, here's what you need to do to

[00:07:28] [SPEAKER_02]: what you put out them and so on and so forth. It sounds simple and indeed it is, but

[00:07:34] [SPEAKER_02]: simple solutions tend to also be very powerful and sometimes simple solutions can also be a

[00:07:40] [SPEAKER_04]: fascinating one of the things that comes immediately to mind is the amount of attention and focus

[00:07:48] [SPEAKER_04]: and investment on matching job descriptions to resumes. And you've got two sort of faulty,

[00:07:57] [SPEAKER_04]: two-dimensional, you know, objects that you're trying to try to match up but to what end. I mean,

[00:08:06] [SPEAKER_04]: to your point, it may sound like a simple approach but it seems much more logical when we think about

[00:08:17] [SPEAKER_04]: quality of higher and metrics that matter, you know, people that are going to come and they're

[00:08:21] [SPEAKER_04]: going to, you know, achieve some semblance of the potential that you try to anticipate and they're

[00:08:28] [SPEAKER_04]: you're going to be able to retain them and keep them engaged and things like that. Like it just

[00:08:33] [SPEAKER_04]: seems like much more logical to invest and, you know, brick in those type of solutions than to

[00:08:40] [SPEAKER_04]: spin your wheels, trying to find a matter match on paper. The reality is that ideally yes,

[00:08:50] [SPEAKER_02]: you would do away with resumes completely but I love the great solution. I also love the solution

[00:08:56] [SPEAKER_02]: that people can apply with today's environment. So yeah, the way I like to think about people and

[00:09:05] [SPEAKER_02]: I love this analogy because it's so terrible is that people are like icebergs. There's a tiny

[00:09:13] [SPEAKER_02]: part that you see that's above the world of surface and that's usually their knowledge, their

[00:09:19] [SPEAKER_02]: experience. Sometimes a bit of their skills to the extent where this can be, you know, depicted in

[00:09:26] [SPEAKER_02]: the resume and think the thing about the enrologen skills is that this stuff is necessary

[00:09:33] [SPEAKER_02]: to the job but it's not sufficient. If you look for an accountant, the fact that I know accounting

[00:09:39] [SPEAKER_02]: doesn't mean that this is enough for me to do the job and by the way, this is the stuff that's

[00:09:45] [SPEAKER_02]: easiest to do so and that's important because we live in a world where the half-life of skills

[00:09:51] [SPEAKER_02]: is rapidly diminishing and used to be five years now people are saying always probably around

[00:09:56] [SPEAKER_02]: half years. So no matter who you hire, you'll end up having to train them. So you know that stuff

[00:10:03] [SPEAKER_02]: that we typically use for selection which is, you know, the A's to set of selection. It's not

[00:10:09] [SPEAKER_02]: that helpful and then it's all this other part of the person, everything that's below the water

[00:10:13] [SPEAKER_02]: line and you're going to find there, you know, they're at huge, there are some not the traits,

[00:10:20] [SPEAKER_02]: there are growth minds at how they're thinking about it, the potential, there are very little

[00:10:25] [SPEAKER_02]: in your things and these are the things, the exact things that are going to determine,

[00:10:31] [SPEAKER_02]: how successful they're going to be in the future. There is a chance of that is let's say that

[00:10:37] [SPEAKER_02]: you know, the biggest correlation between one-weeking measure and success in the future job

[00:10:41] [SPEAKER_02]: is actually those things and more importantly, these are the things that are most difficult to change.

[00:10:48] [SPEAKER_02]: Everyone, including myself, I've done a lot of work on myself for very few years right

[00:10:52] [SPEAKER_02]: and I've only managed to change that part under the water, all the tiny changes.

[00:11:00] [SPEAKER_02]: I've learned how to come with lies, you know, and maybe show different things on top. But really,

[00:11:05] [SPEAKER_02]: I've never loved rules. I still hate rules. I'm just not wise enough to know when to

[00:11:12] [SPEAKER_02]: conform to it, but I'm never going to be a number than whenever you're going to put me in a position

[00:11:17] [SPEAKER_02]: where I have stripped for the rules. I'm going to have a hard time dealing with that. So, once you start

[00:11:23] [SPEAKER_02]: thinking about that and you're saying hey, what are you going to hire? You want to hire that

[00:11:28] [SPEAKER_02]: tiny piece and where a lot of people look alike because everyone's the same and you know they all

[00:11:33] [SPEAKER_02]: know at an Java or Python or everybody has done a bit of AI. Or do I care about this icebergs?

[00:11:41] [SPEAKER_02]: And those are some of them are small, some of them are big, some of them are actually some

[00:11:45] [SPEAKER_02]: of them are huge depending on what I'm looking for. There is a very big difference that's

[00:11:51] [SPEAKER_02]: you really can tell. So, thinking that way is a great way to start thinking about hiring.

[00:11:58] [SPEAKER_02]: And yes, you know, the resume is going to be with us for a foreseeable future. Hopefully there's

[00:12:03] [SPEAKER_02]: not some better alternative yet. Everybody's trying their hand at having a blockchain about skills.

[00:12:09] [SPEAKER_01]: It's an interesting concept. If you like swiping, then head over to Substack and search

[00:12:14] [SPEAKER_01]: or work to find. WRK, to find and subscribe to the weekly newsletter.

[00:12:20] [SPEAKER_02]: But until that happens, sure, it's going to be an factor of making hiring decisions.

[00:12:28] [SPEAKER_02]: And this kind of drives to what the heck do we have challenges? Because we don't want to just make

[00:12:34] [SPEAKER_02]: decisions based on a couple of data coins that you can put in a different paper.

[00:12:40] [SPEAKER_02]: I want to must all the information I get about first. Yes, I want to know about

[00:12:45] [SPEAKER_02]: their skills. Yes, I want to know about the aspirations. Yes, I want to know about their

[00:12:50] [SPEAKER_02]: second-medics and who they are now. And the more data I have about them, the better my chance

[00:12:55] [SPEAKER_02]: of making sure that I'm maximum to the right, geek, role, occupation, career, all these things.

[00:13:03] [SPEAKER_02]: It's kind of the reason why they're career class looks, it's cool right. They have a certain function.

[00:13:09] [SPEAKER_02]: And we've been very negligent and provided in the same function for employees.

[00:13:15] [SPEAKER_04]: Well, certainly there's just cases for this on the candidates side and the pretty higher side as

[00:13:19] [SPEAKER_04]: well as post-higher, it's a metallic management perspective. I mean, I also think this is valuable.

[00:13:28] [SPEAKER_04]: I'm not saying that all psychometric assessments are necessarily equal, but it does seem like

[00:13:36] [SPEAKER_04]: if these are attributes that don't necessarily change, it would make sense to have that as part of

[00:13:46] [SPEAKER_04]: your portable credentials, whatever you plan to do, right? It could be applying for a job.

[00:13:55] [SPEAKER_04]: It could be applying to graduate school. It could be, I don't know, it just seems like again,

[00:14:00] [SPEAKER_04]: as long as you have control over who accesses that information and you assume that you

[00:14:07] [SPEAKER_04]: agreed with it, and that's how it wound up there. That would make the application process in

[00:14:12] [SPEAKER_04]: the future a lot easier. I mean, you could assess for all of those true, you know, human

[00:14:18] [SPEAKER_04]: skills which at the end of the day, like you said, as technical skills, the half-life

[00:14:25] [SPEAKER_04]: strengths and AI continues to take on more, you know, a sort of upscale itself in a way.

[00:14:31] [SPEAKER_02]: Oh, I wish there was a super bullet, but it's really interesting that you're saying this because

[00:14:35] [SPEAKER_02]: we were working now at the major university and we were working them both ways. The first part

[00:14:42] [SPEAKER_02]: is hey, two people come in and they start their studies and we want to give them some directional advice.

[00:14:50] [SPEAKER_02]: It's not so much study that we come in the country, but it's like, look, here are areas of interest

[00:14:55] [SPEAKER_02]: that are going to match your personality. Here are things that you're going to enjoy doing over

[00:15:00] [SPEAKER_02]: the next couple of years. And by the way, these are potential, you know, a few patients that come

[00:15:05] [SPEAKER_02]: out of that. It's not funding, it's necessarily a one-to-one, you know, I'm going to pick, oh, I want

[00:15:09] [SPEAKER_02]: to become a dentist, but it's more like, you know, whether they enjoy interacting with people and

[00:15:15] [SPEAKER_02]: more of, if you've ever done the following some career, professional theory. It's really great.

[00:15:21] [SPEAKER_02]: It looks like some architects and it doesn't necessarily describe everyone and everything,

[00:15:27] [SPEAKER_02]: but it kind of gives you a rough idea of where you want to be. That's great and you do that when

[00:15:31] [SPEAKER_02]: you enter a school and you spend a couple of years learning about these things. And by the time

[00:15:37] [SPEAKER_02]: you get out, no matter what you've started, now it's the time where you're saying hey, okay,

[00:15:41] [SPEAKER_02]: I have this skills. This is the first time I am, where are the kind of things I can do now?

[00:15:47] [SPEAKER_02]: And then you match them to the job market available. And I think that over time we're going to be

[00:15:52] [SPEAKER_02]: seeing more and more people thinking that way. Let's help people in both hands, right? Let's

[00:15:56] [SPEAKER_02]: give them their interaction while they need to study and let's make sure when they get out.

[00:16:01] [SPEAKER_02]: They also get much better at kind of things because I can start it becoming an architect

[00:16:06] [SPEAKER_02]: and I can be an architect or I can run an architectural office. Those are both occupations that seem

[00:16:12] [SPEAKER_02]: the same, but nothing alike. And you know, I can become an editor for architecture because

[00:16:19] [SPEAKER_02]: I love architecture, but I hate practicing it and I love teaching about these are the kind of things

[00:16:24] [SPEAKER_02]: that sound simple and principal. But once you start and say hey, the data says that you're an architect,

[00:16:31] [SPEAKER_02]: you've done it, you love your school. You're going to hate being an architect and you know what

[00:16:37] [SPEAKER_02]: this is here, if you can't do something, it's about it. But it's a very viable career choice.

[00:16:44] [SPEAKER_02]: I think we're going to get into a nature where career choices are going to be more informed

[00:16:50] [SPEAKER_04]: and people that are going to be way more successful. I think you can give that guidance after age 16.

[00:16:57] [SPEAKER_04]: I think people would be happier. In high school, so my daughter will be a junior,

[00:17:04] [SPEAKER_04]: in high school next year and she'll start looking at colleges and taking SATs and things like that.

[00:17:09] [SPEAKER_04]: So start thinking about where she wants to go and what her major might be or whatever. If your

[00:17:14] [SPEAKER_04]: school policy is do your own work. And to the assist of embracing AI for good as a tutor, as a coach,

[00:17:25] [SPEAKER_04]: not cheat. Like the jobs that might be appealing to them now and why they want to go to a

[00:17:32] [SPEAKER_04]: certain school or going to a certain program, you don't know what the longevity of that

[00:17:38] [SPEAKER_04]: career path really is. I mean already we have its spending hundreds of thousands of dollars on

[00:17:44] [SPEAKER_04]: a higher education and majoring in things that have no short-term ROI to

[00:17:53] [SPEAKER_04]: coupe that cost. I know we can't plan too far ahead because we never know how things get

[00:18:02] [SPEAKER_04]: going to evolve but we can sort of plot multiple trajectories of where AI might go and the

[00:18:10] [SPEAKER_04]: capabilities that it may take on, unless you're the absolute best at doing that, most of those roles

[00:18:22] [SPEAKER_02]: well done by digital labor in some way. You raise a very interesting point and we expect AI to fix

[00:18:29] [SPEAKER_02]: all sorts of broken processes but a lot of times when I talk about AI, I find the introduction of AI

[00:18:37] [SPEAKER_02]: this days is very analogous to the introduction of computers in the first place. We used to have

[00:18:43] [SPEAKER_02]: computers, they got everything that has been written as commercially and people always think

[00:18:49] [SPEAKER_02]: the AI game was kind of like the widespread computer that was in the late 70s early 80s where

[00:18:55] [SPEAKER_02]: the modern PC era generated and I guess what we had exactly the same reactions. We're going to

[00:19:03] [SPEAKER_02]: take away their jobs, no one's going to be able to find a job now computers are going to do everything

[00:19:08] [SPEAKER_02]: and then it was like okay we can take everything and automate it and what we did in that process

[00:19:15] [SPEAKER_02]: is how we learned is that we took a lot of broken processes, things that didn't work in the

[00:19:21] [SPEAKER_02]: first place. We computerized them and guess what? Very main broken. Now it was just that they were

[00:19:28] [SPEAKER_02]: breaking faster and we're getting their own results faster and the same thing applies with AI

[00:19:34] [SPEAKER_02]: it's not going to solve problems by itself you have to know how to solve the problem

[00:19:39] [SPEAKER_02]: you have to know the data, you can of course get tremendous insight much like computers allowed us

[00:19:44] [SPEAKER_02]: no bigger sets of data and do data processing in ways we didn't imagine for AI allows us to do the same

[00:19:51] [SPEAKER_02]: thing but if I just take AI and finish the data that I have today especially the hiring area

[00:19:58] [SPEAKER_02]: if I take an AI say hey look at all my hiring decisions these are the good ones these are the

[00:20:02] [SPEAKER_02]: problems replicate those it's going to be fantastic job in replicating them which means that

[00:20:08] [SPEAKER_02]: it's going to replicate all the buyers it's going to replicate all the inequalities and

[00:20:14] [SPEAKER_02]: I have yet to meet a company where I can say that every hiring decision has been a good one

[00:20:19] [SPEAKER_02]: everything that's upstander in the base companies and the bad way that we have about managed performance

[00:20:23] [SPEAKER_02]: and all those things they're going to be replicated perfection so that's not the right way to do it

[00:20:29] [SPEAKER_02]: we should always go take a step back and say okay now I have a tool that's way more powerful than before

[00:20:35] [SPEAKER_02]: and this tool now is going to computer know this now it is an AI how am I going to solve this

[00:20:41] [SPEAKER_02]: let's not just apply things to the existing processes the biggest step back and think about

[00:20:48] [SPEAKER_02]: how do we engineer the process is the optimal process what should it look like and we've done this

[00:20:54] [SPEAKER_02]: in a couple of places with AI and a lot of times when we take we we're using some a bunch of

[00:21:00] [SPEAKER_02]: LLMs in our current limitations and we're never letting the LLM do everything because once you do that

[00:21:08] [SPEAKER_02]: then the LLM makes all sorts of shortcuts and inferences and whatnot that have no way of

[00:21:14] [SPEAKER_02]: none but the LLM is great and fantastic in understanding text and extracting meaning

[00:21:20] [SPEAKER_02]: and doing all those things and if you give it to me then you do it to this by a person you can control

[00:21:26] [SPEAKER_02]: what what was from one place to the other it's the best tool we've had in years I mean we're

[00:21:32] [SPEAKER_02]: we're sharing out software like never before with the better quality than ever before

[00:21:36] [SPEAKER_02]: with the lower cost than ever before let's say I do think probably the most appropriate

[00:21:43] [SPEAKER_04]: analog to this era going back to the point you made at the beginning I think just really

[00:21:50] [SPEAKER_04]: understanding how you can derive insights from that data and using AI to sort of augment

[00:21:57] [SPEAKER_04]: our own human intellect and capabilities in terms of you know

[00:22:04] [SPEAKER_04]: gaubleating insight from but information I mean it's I'd like to see us move more to the

[00:22:10] [SPEAKER_04]: augmentation rather than automation in the sense that I think that that's a lot where a lot more

[00:22:17] [SPEAKER_04]: value will be realized and you know as we're talking about hiring and telling acquisition

[00:22:22] [SPEAKER_04]: about making better decisions. As long as you have a good foundation in quality data and you've

[00:22:32] [SPEAKER_04]: assessed that data for potential human bias in that historical data at least you know that

[00:22:39] [SPEAKER_04]: and you can then of course correct and you can try to mitigate you know bias going forward

[00:22:44] [SPEAKER_04]: but to just absorb historical information and then just use that at scale you know

[00:22:52] [SPEAKER_04]: just perpetuating that historical bias that has existed and we've seen unfortunately we've seen

[00:22:58] [SPEAKER_04]: headlines that talk about exact scenario so I guess one of the things that you know I've

[00:23:06] [SPEAKER_04]: talked about before is like as you train these these models when you're looking at

[00:23:12] [SPEAKER_04]: candidate application or employee record how do you make sure that you are looking at it in

[00:23:20] [SPEAKER_04]: such a way like how do we make sure that those biases are in fact mitigated and I know

[00:23:27] [SPEAKER_04]: you know brick has gone through an independent audit to check for an adverse impact

[00:23:33] [SPEAKER_04]: before you delivered it to customers but I wonder like what happens when the

[00:23:39] [SPEAKER_04]: when your client gets hit and they say well okay this is this is awesome and we now have it trained

[00:23:46] [SPEAKER_04]: additionally on our proprietary management data or whatever other you know data they have

[00:23:54] [SPEAKER_04]: in their HR systems but how do you think about that how do you how do you talk to your

[00:24:01] [SPEAKER_04]: customers about maintaining you know that you know that audit posture I want to take a break real quick

[00:24:09] [SPEAKER_01]: just to let you know about a new show we've just added to the network up next at work

[00:24:15] [SPEAKER_01]: hosted by Gene and K to Kiel of the Devon group and Tastic Show if you're looking for something

[00:24:22] [SPEAKER_01]: that pushes the norm pushes the boundaries has some really spirited conversations

[00:24:28] [SPEAKER_01]: Google up next at work Gene and K to Kiel from the Devon group. You know what you should know

[00:24:39] [SPEAKER_05]: you should know that you should know podcasts that's what you should know

[00:24:44] [SPEAKER_05]: because then you'd be in the know on all things that are timely and topical. Subscribe to the

[00:24:50] [SPEAKER_02]: you should know podcasts thanks. There's a laptop back into this statement sample where do we

[00:24:57] [SPEAKER_02]: begin? A very good question thank you for asking. Let's look at a couple of different levels

[00:25:05] [SPEAKER_02]: let's use hiring as an example though but we're going to discuss actually applies to a whole bunch

[00:25:10] [SPEAKER_02]: different areas you talk about the New York City law but when we first saw him said

[00:25:16] [SPEAKER_02]: oh why does the civil exist this is super close? He or she has been saying here's the four

[00:25:21] [SPEAKER_02]: fifth room you have to abide by since probably something in the 1970s you know right there

[00:25:28] [SPEAKER_02]: around Title VII and whatnot yeah and then with all about it a bit and we said no it's always been

[00:25:36] [SPEAKER_02]: very cumbersome for people to actually abide by the course we saw or even have proper reporting.

[00:25:42] [SPEAKER_02]: Now with technology we shouldn't actually relax and say you know we've been doing him a half

[00:25:47] [SPEAKER_02]: us job manually that's good thing you're doing a half us job with AI no AI is actually an opportunity

[00:25:55] [SPEAKER_02]: to become a better version of ourselves and yes the New York City law makes sense because they say

[00:26:01] [SPEAKER_02]: you're using a system of crunches they feel like there's nothing more of take a moment

[00:26:05] [SPEAKER_02]: and make sure that when you're crashing this data you don't have anything that kind of breaks

[00:26:11] [SPEAKER_02]: your system and it reduces bias and anybody who's written software that's more than two lines

[00:26:16] [SPEAKER_02]: knows that you can introduce a lot of bias to design errors thinking errors actual coding errors

[00:26:23] [SPEAKER_02]: data errors there are a million ways where things go wrong so having a third part in that's part

[00:26:30] [SPEAKER_02]: of the nice thing because people are always in policing that say hey you're playing with people

[00:26:35] [SPEAKER_02]: scared let's have a third part there are a bunch of people who can build this and you're one

[00:26:41] [SPEAKER_02]: of those where you say hey we love break it's great now but look at the data and make sure

[00:26:46] [SPEAKER_02]: the taxid does what it says let's make sure we're doing right by candidates and by employees

[00:26:51] [SPEAKER_02]: whenever we make an employ of decision this is slightly utopic and it's not going to start working

[00:26:58] [SPEAKER_02]: from day one it's going to be a long, long time before this things actually get traction people

[00:27:04] [SPEAKER_02]: start using them all the time but it's definitely step into their direction because that's what

[00:27:09] [SPEAKER_02]: you want to say whether you're an employee air and an employee a candidate or anywhere in between

[00:27:16] [SPEAKER_02]: you want to know that you're going to go through a system and the system is going to judge you

[00:27:20] [SPEAKER_02]: on married equitable and it's going to be not going to be impacted by how old you are where you were born

[00:27:27] [SPEAKER_02]: all those things so I love it it's painful and that's what happens with change right all of a

[00:27:34] [SPEAKER_02]: sudden you have new technology and we're very coming to terms with what we're doing and how this

[00:27:39] [SPEAKER_02]: things working and then all of a sudden you have legislation and I know very ship people who

[00:27:46] [SPEAKER_02]: have legislation and then definitely not one of them even though in principle I am a green I think

[00:27:52] [SPEAKER_02]: it's the right thing to do do I experience the pain yes doesn't slow the pace of innovation yes

[00:27:59] [SPEAKER_02]: does it also make sure that think about a farmer industry without any controls without

[00:28:05] [SPEAKER_02]: doing cultural rights I don't want to live in that world why do we think the fact that this

[00:28:10] [SPEAKER_02]: is just a lot much less than that so yes you know striking the right balance between

[00:28:16] [SPEAKER_02]: responsible using AI and ways that people understand make sure that there are third party or

[00:28:21] [SPEAKER_02]: it's the folder solar accountable and I don't know you seem to be EUA AI art it's a whole new chapter

[00:28:30] [SPEAKER_02]: of regulation legislation and commitments it's a lot it's not unreasonable and that's the challenge

[00:28:39] [SPEAKER_02]: there is a part of me that says I want to go ahead and innovate and I'll do it responsibly

[00:28:45] [SPEAKER_02]: but how can I guarantee that everybody is going to be responsible so maybe it's a good thing

[00:28:52] [SPEAKER_02]: maybe it's a playing field in even playing field for everybody making sure that yes we all do AI

[00:28:58] [SPEAKER_02]: but yes we are not negligent in what we have to do in on doting our eyes and crossing our

[00:29:05] [SPEAKER_02]: pace and making sure that we don't think so well those are my short answer do you know want the long

[00:29:09] [SPEAKER_04]: answer you know I've read sort of the highlights of the EUA act it's certainly very different

[00:29:17] [SPEAKER_04]: than the legislation that exists in New York City but I can see that some states in the US

[00:29:27] [SPEAKER_04]: already you know sort of taking cues from from the EUA act frankly I mean if you're

[00:29:33] [SPEAKER_04]: a global organization or you have ambitions to be one you've kind of got to pay attention to

[00:29:41] [SPEAKER_04]: the EUA act not because you have aspirations to enter European markets but because that's

[00:29:49] [SPEAKER_04]: probably going to be one of the sort of benchmarks in terms of legislation that other people

[00:29:56] [SPEAKER_04]: are going to continue to sort of align themselves with from a from a risk standpoint and from a

[00:30:05] [SPEAKER_04]: liability standpoint in terms of who's who's responsible other areas may not have the same

[00:30:11] [SPEAKER_04]: penalties and they may not have necessarily like a GDPR type of you know data privacy you know

[00:30:19] [SPEAKER_04]: pre-existing legislation or this A.I.E.A. act has sort of be a layer on top of that but I still

[00:30:26] [SPEAKER_04]: think if you can comply with the EUA act you can probably comply with most other you know legislation

[00:30:34] [SPEAKER_04]: as it you know propagates around the world so we'll see how people do against that but I do think

[00:30:42] [SPEAKER_04]: you know data privacy is important maybe that plays into the you know the blockchain

[00:30:46] [SPEAKER_04]: you know business case where visuals have more more control over their or their professional

[00:30:53] [SPEAKER_04]: data but I think people at least in here in the US we were still trying to get their arms around

[00:31:00] [SPEAKER_04]: that and to your point from New York so these standpoint it's not most enforceable pieces

[00:31:06] [SPEAKER_04]: legislation so companies aren't exactly you know beating down my door to you know get out of it

[00:31:13] [SPEAKER_02]: look you know how this legislation started people are ignoring them until the first couple of instances

[00:31:19] [SPEAKER_02]: of enforcement come like yeah and then everybody rushes through to fix them is there I think to

[00:31:27] [SPEAKER_02]: do yes is it the best measure no but I don't know what the best measure would be at this point

[00:31:34] [SPEAKER_02]: but it's a new game right both companies and legislation they're trying to cut out who

[00:31:42] [SPEAKER_04]: agree out of this so it's interesting times we'll see whether the US federal government does because

[00:31:49] [SPEAKER_04]: they've been I know they've been having meetings and they've said committees or whatever but we

[00:31:53] [SPEAKER_04]: know that in itself doesn't mean that action is being taken but I think everyone needs to say on

[00:32:01] [SPEAKER_04]: photos and take a human centric approach to to all this we're talking about people's livelihoods

[00:32:07] [SPEAKER_04]: and I do think these things are important when it comes to innovation I mean I think some

[00:32:13] [SPEAKER_04]: of the legislation is a little bit misguided in terms of where they're trying to put some of

[00:32:18] [SPEAKER_04]: their restrictions and the AI field in general fan of open source in general and in this case I

[00:32:25] [SPEAKER_04]: think they open source models or more transparent but they can also cost companies a lot more

[00:32:32] [SPEAKER_04]: money if they need to build things you know themselves as a cost of compute resources is still

[00:32:39] [SPEAKER_04]: pretty high but you know I'm more opponent of legislating at the application level or the use

[00:32:45] [SPEAKER_04]: case level that I am about trying to control you know only these upper echelon of you know LLM providers

[00:32:53] [SPEAKER_04]: you know setting rules based on the amount of compute that they're using or these metrics that

[00:32:59] [SPEAKER_04]: won't apply to 99.5% of companies on the planet. Hi there I'm Peter Zolman I'm a co-host of the

[00:33:07] [SPEAKER_06]: inside job boards and recruitment marketplaces podcast and I'm Stephen Rothberg and I guess that

[00:33:12] [SPEAKER_06]: makes me the other co-host every other week we're joined by guests from the world's leading job site

[00:33:18] [SPEAKER_03]: together we analyze news about general niche and aggregator job board and recruitment marketplaces sites.

[00:33:24] [SPEAKER_02]: Make sure you sign up and subscribe today. We've always seen two schools of regulation and I think

[00:33:32] [SPEAKER_02]: the US and the EU are quite the two holes in comparison with the US always tries to

[00:33:40] [SPEAKER_02]: greatly with controls. I think about surveying soaps here I have to do this and check that

[00:33:45] [SPEAKER_02]: level of law and so on and measure this and measure that in New York City show me your bias

[00:33:50] [SPEAKER_02]: and it's fine and the European more of the liberal minded Europeans that they are

[00:33:58] [SPEAKER_02]: they always like to regulate by almost by intent here are the things that you should do

[00:34:04] [SPEAKER_02]: and here's how you should be structured you shouldn't do this but GDPR has been around now for

[00:34:11] [SPEAKER_02]: must be almost a decade and there is still now it's emerging in our nice standard that's

[00:34:18] [SPEAKER_02]: corollary to GDPR but there's never been a GDPR situation and that's one way to regulate right

[00:34:24] [SPEAKER_02]: saying hey can I think that we should do and if I find you not doing those things

[00:34:28] [SPEAKER_02]: they're going to be paying off this debate but you don't have to prove all the time that you're doing

[00:34:31] [SPEAKER_02]: them and but you have to provide them for the other thing which is show me the controls and

[00:34:39] [SPEAKER_02]: you find the way to bypass the whatever it is that I meant but your controls to look fine

[00:34:44] [SPEAKER_02]: it's kind of okay frankly in the little bit of hope and there are no good solutions

[00:34:50] [SPEAKER_02]: and legislation has all been historically been very slow cuts up with new technologies

[00:34:56] [SPEAKER_02]: these days technologies slow cuts up with AI so I think of myself a very technical minded person

[00:35:04] [SPEAKER_02]: having to background with your science and all that stuff and on occasion I am a little

[00:35:09] [SPEAKER_02]: well than like natural networks or what's implication of that where does my data actually go

[00:35:15] [SPEAKER_04]: from here I have questions so when you look at the broader space in terms of energy AI

[00:35:24] [SPEAKER_04]: outside of you know talent acquisition outside of HR and talent believe the

[00:35:31] [SPEAKER_04]: tools that you see that either you know scare you or I think this is fantastic whether

[00:35:38] [SPEAKER_02]: that's work related or in your personal life I'm excited and I don't tend to be scared I see

[00:35:47] [SPEAKER_02]: a lot of wonderful opportunities look like any tool we can use a hammer to build things

[00:35:54] [SPEAKER_02]: you can use a hammer to smash your head it's never the hammer's fault right first that's

[00:36:02] [SPEAKER_02]: much in my head it would be stupid and why would you do it but sometimes it's not as clear

[00:36:08] [SPEAKER_02]: or not as painful in me getting so we've been I'll I'll focus first about an example on our

[00:36:17] [SPEAKER_02]: place where we've been doing talent failures there's always they need to have a pulse system

[00:36:23] [SPEAKER_02]: like Sonom you're in apology whatever you call it with all the hearts you know you have to

[00:36:28] [SPEAKER_02]: know this and you know it's like computer science and then there is programming and then there's

[00:36:33] [SPEAKER_02]: line of distance on but at the time you build this you've spent like six months you've cut all

[00:36:38] [SPEAKER_02]: of everything and you know you put the final dog and you're saying oh look at what we've done

[00:36:44] [SPEAKER_02]: now this is all outdated because it's been six months that will be doing this and they're like five

[00:36:49] [SPEAKER_02]: machines here and hand there and those things are obsolete and that was really hard master

[00:36:55] [SPEAKER_02]: nowadays with a help of federal lands with build systems that are dealing with skills and they don't

[00:37:02] [SPEAKER_02]: need a firm taxonomy they kind of rearrange themselves and they're not from complex and they can

[00:37:07] [SPEAKER_02]: compare skills without them needing to be one dot two dot seventeen or whatever the owner of

[00:37:13] [SPEAKER_02]: because by the way I still look very useful input but you're giving this level of flexibility where

[00:37:20] [SPEAKER_02]: we're starting to become better at handling slight more ambiguous data. Translation is one of the

[00:37:26] [SPEAKER_02]: areas where we've seen hugely profound every time I go into a call with a customer partner one of my colleagues

[00:37:34] [SPEAKER_02]: I get a fantastic transcript out of it which is great voice recognition and then I get an even better

[00:37:42] [SPEAKER_02]: but captures the essence of what I said that would be a note taker six months ago

[00:37:48] [SPEAKER_02]: I'm not even going to talk about the years so all sorts of response to things are happening and

[00:37:55] [SPEAKER_02]: people are complaining because they're saying oh you know this new day in customer we're doing

[00:37:59] [SPEAKER_02]: they're now no longer paid that well and that's a downside people are saying they're like a good

[00:38:03] [SPEAKER_02]: threatened at the same time the same people are now saying hey you know what I could do this in

[00:38:10] [SPEAKER_02]: a year now can do it in four months because I'm longer half of this meeting task and I have the

[00:38:16] [SPEAKER_02]: systems that can do it for me and you know we have a container in content and the content right

[00:38:21] [SPEAKER_02]: is very concerned because all the areas going to generate all the content it can refine a lot

[00:38:26] [SPEAKER_02]: of it but you still need to generate some original thoughts and that's I have to get to see the

[00:38:32] [SPEAKER_02]: other members from a temporary period so all I'm saying that it's going to be given taker something

[00:38:38] [SPEAKER_02]: going to get water something's going to go away many things are going to come up and that's the

[00:38:44] [SPEAKER_02]: exciting part because I think we've been doing things separate for ages and I'm the field of

[00:38:51] [SPEAKER_02]: HR so if I look back at HR we're doing things pretty much the same way that we're doing it

[00:38:57] [SPEAKER_02]: for a hundred years ago because there wasn't a clear actionable better way to do them now I

[00:39:04] [SPEAKER_02]: give this better ways to do things and that's going to happen in a number of foods and that's

[00:39:09] [SPEAKER_02]: going to be disruptive I like to be a technical optimist and say that that's going to be all

[00:39:13] [SPEAKER_04]: they're all a positive thing. I'm definitely in your camp on that but I think people need to

[00:39:19] [SPEAKER_04]: accept that this is this is a significant change that this is a transformation that is

[00:39:26] [SPEAKER_04]: broader and more complex but also perhaps has the most sort of net positive for them

[00:39:33] [SPEAKER_04]: to visually for their team and hopefully for their organization but yes you've got to adopt

[00:39:41] [SPEAKER_04]: these tools that you talked about the beginning about adoption I think that's getting people to actually

[00:39:47] [SPEAKER_04]: use these tools and be comfortable interacting with these tools I think is

[00:39:53] [SPEAKER_04]: it's very important to be able to do so responsibly. The average person is right now in early

[00:40:02] [SPEAKER_04]: adopter or user of these tools but as we know you can now create your own age and your own

[00:40:13] [SPEAKER_04]: hope pilots now so raw builders as well and when I think about how be sort of up skill everyone

[00:40:21] [SPEAKER_04]: regardless of raw or seniority it's easy to look around read the headlines and think this is

[00:40:29] [SPEAKER_04]: overwhelming I'm not sure I can handle all this or I'm not sure how many days till I retire

[00:40:36] [SPEAKER_04]: like do I need to go through this but I think like you said it's just like learning how to use

[00:40:41] [SPEAKER_04]: a computer you know we don't stop using a computer when you retire I mean my parents are in

[00:40:47] [SPEAKER_04]: early 80s there's a test to their devices as mere even my daughter I think so this is a life

[00:40:56] [SPEAKER_04]: you know adjustment I think that people need to get comfortable with but they can't lose

[00:41:01] [SPEAKER_04]: they can't just offload everything too AI they still need to think critically about

[00:41:07] [SPEAKER_04]: how they're using it as well as you know what the what the output is I mean it's still not a doctor

[00:41:13] [SPEAKER_04]: so be smart about the types of questions that you're asking it but also understand that

[00:41:20] [SPEAKER_04]: it's not giving you a definitive and 100% accurate answer and so I think that's part of what

[00:41:27] [SPEAKER_02]: people need to go into to go back to my hammer analogy everybody gets a brand new set of AI

[00:41:33] [SPEAKER_04]: hammers how they're gonna use them is up for them that of course brings me to the one question

[00:41:40] [SPEAKER_04]: I asked oh my guests because which is when you see the phrase elevator AICU comes to mind

[00:41:49] [SPEAKER_02]: I thought about it for a bit and you know I can I spent the best decade or so

[00:41:56] [SPEAKER_02]: negative or maybe even a bit further in psychometrics and I'm still not a psychometration by far

[00:42:03] [SPEAKER_02]: but the AICI rectal thing of the AICU not so much as a measure of clever one is but as a

[00:42:11] [SPEAKER_02]: measure of cognitive ability how can people handle specific tasks and when you're talking about

[00:42:19] [SPEAKER_02]: AICU what immediately jumps to my mind is how can people handle the AICU but so I've

[00:42:25] [SPEAKER_02]: really deep to how can handle the AICI but so I've really deep to deal with AICI and if I have a

[00:42:32] [SPEAKER_02]: AICU and that's my very little interpretation of your question now I understand that

[00:42:38] [SPEAKER_02]: fantastic too I'm gonna use it to validate things I'm gonna use it to help me unstuck when I'm

[00:42:45] [SPEAKER_02]: gonna take a lot of tedious work I was feeling the other day some compliance questioner

[00:42:51] [SPEAKER_02]: yes I love the fact that I can have AICI help with that you know giving the answer and just

[00:42:55] [SPEAKER_02]: review them I still don't trust the AICI for AICI answers from me I would never give up any compliance

[00:43:01] [SPEAKER_02]: person or anything else for that matter is just been done by an AICI so does it take 80% of the

[00:43:07] [SPEAKER_02]: work for me oh yeah can I use the right tool in the right place and in the right mode but you're

[00:43:14] [SPEAKER_02]: a is it easy to get I think it's challenging but hey we can all work on it that's absolutely right

[00:43:24] [SPEAKER_04]: that's absolutely right now I love it I love you take more because as always pleasure to talk to you

[00:43:30] [SPEAKER_04]: thank you so much for spending some time with me and with our audience I think there's a lot of

[00:43:35] [SPEAKER_02]: AICI ways from the conversation thank you for having me you know that this is one of my favorite

[00:43:39] [SPEAKER_02]: subjects and I'm sorry we didn't have more time I'd love to expand on any number of items

[00:43:44] [SPEAKER_02]: and it's really interesting every time I talk about AI even this conversation I learned something new

[00:43:52] [SPEAKER_02]: and it helps me think again about the same items over and over and become a little bit better in

[00:44:00] [SPEAKER_02]: you know what I think I've raised my AI to it today that's that's the sound by right there

[00:44:07] [SPEAKER_04]: thank you Mark Ellis always a pleasure thank you both thank you everyone for this thing I

[00:44:11] [SPEAKER_04]: wraps up another episode of LVature AICI Q see you next time thank you I get it the podcast just isn't

[00:44:32] [SPEAKER_01]: enough that's all I had over to your favorite social app search a work to find WRK defined and connect with us