Ep 35: A Whirlwind 360-Degree Tour of AI in HR with Lydia Wu

[00:00:00] Welcome to Elevate Your AIQ, the podcast focused on the AI-powered yet human-centric future

[00:00:04] of work. Are you and your organization prepared? If not, let's get there together. The show

[00:00:09] is open to sponsorships from forward-thinking brands who are fellow advocates for responsible

[00:00:13] AI literacy and AI skills development to help ensure no individuals or organizations

[00:00:17] are left behind. I also facilitate expert panels, interviews, and offer advisory services

[00:00:22] to help shape your responsible AI journey. Go to ElevateYourAIQ.com to find out more.

[00:00:39] Thanks for listening to another episode of Elevate Your AIQ. Today I'm thrilled to sit

[00:00:43] down with Lydia Wu, a well-known expert in all things HR tech, talent acquisition, skills,

[00:00:48] people analytics, AI, you name it. I've been a follower of Lydia's for quite a while, not

[00:00:53] only for her expertise, but also she doesn't pull any punches when assessing solutions and

[00:00:57] the market more broadly. You should definitely check out her webinars and multimedia content,

[00:01:01] particularly her new Oops! Did I Think That Out Loud series. I'll put some links in the

[00:01:06] show notes. In this wide-ranging conversation, we'll dive deep into everything from the critical

[00:01:10] need for transparency and responsible AI practices to the exciting possibilities that generative

[00:01:15] AI tools are bringing to HR and talent research. Whether you're an HR professional navigating

[00:01:20] AI-driven transformation or somebody curious about how this technology is reshaping the future

[00:01:25] of work, you're in for a treat. Hello everyone, welcome to another episode of Elevate Your AIQ.

[00:01:31] I'm your host, Bob Pulver. Today I have the pleasure of speaking with Lydia Wu. How are you, Lydia?

[00:01:36] Doing very well. Thanks for having me here. Absolutely. I appreciate you being here.

[00:01:40] Just to kick things off, why don't you give the listeners just a little bit of detail around

[00:01:45] your background and what you're working on? Yeah, absolutely. So nice to meet everyone.

[00:01:49] Thanks for listening to this episode. My name is Lydia and I've been in HR tech for about 15 years.

[00:01:53] At this point, I call myself the full 360 of HR tech. So I started out as a consultant

[00:01:59] implementing HR technology. Went in-house to run an HR tech portfolio, HR tech buying strategy,

[00:02:05] which included people analytics and a little bit of AI as well. Got a little bored of that,

[00:02:09] so pivoted out, went on to the solution provider side. And now I'm working with a variety of startups

[00:02:16] in terms of what it's like launching a startup in the HR tech world. What's the ecosystem like?

[00:02:21] How do you create a solution that's best fit for the end users? And simultaneously,

[00:02:26] I'm also running my little channel called Whoops, Did I Think That Out Loud?

[00:02:30] Yeah, I definitely want to hear a little bit more about that. So we were just talking in the green room

[00:02:37] about your Galileo review, Josh Burson's review. Is that the kind of thing, we'll get into that as

[00:02:42] we talk, but is that the kind of thing that you're going to be doing? Just taking a firsthand look at

[00:02:50] some of these tools and calling people out if they're doing some not so great things?

[00:02:56] BSing? Yes. Absolutely. So I think the whole premise of How Whoops Did I Think That Out Loud even came

[00:03:02] about was, I distinctively recall sitting in a meeting room, and this was a pretty senior meeting

[00:03:07] room. And I'm listening to all these HR executives talk. And there was a moment where I'm like,

[00:03:11] oh my God, the stuff that you say in this room can never exit this room. Do people you work with

[00:03:17] actually know that this is what you think? And this is like what's going through your head as

[00:03:21] you're kind of going out there and championing it's about the people, it's about like the engagement,

[00:03:26] et cetera, et cetera. So that was when I kind of started that, I think that I allowed the very

[00:03:30] first article, which is about DEI in the corporate world, essentially. And since then, I've just strived to

[00:03:36] be sort of like the real talk BFF, as I call myself in HR tech, which is essentially like,

[00:03:42] let's call things for what they are. If it's legit, let's say like, yep, it's legit, move on. If it's

[00:03:47] not legit, give some positive feedback. There's no need to chase after people with pitchforks, but there's

[00:03:52] also, I don't think, room for snake oil necessarily to be sold in the market anymore.

[00:03:58] Yeah, it's funny. I guess I kind of do that off the record, like whether it's Monday morning

[00:04:04] and quarterbacking, as they say, or what have you. But this whole show is around responsible AI,

[00:04:11] and I'm always pushing people to think more deeply about the things that they're building,

[00:04:15] as well as just this concept of being responsible by design, just like we used to say,

[00:04:21] privacy by design and all these things. If you think about it from the beginning,

[00:04:25] from conception, then you have less to worry about. And you can, I guess, move fast,

[00:04:33] maybe break a few things, but make sure that you're not completely propagating these behaviors.

[00:04:42] And we've seen that a lot with some of the generative AI solutions, certainly.

[00:04:47] For sure. And I think this whole notion of moving fast and breaking things in the world of HR tech,

[00:04:51] it's almost like we took the surface without understanding the deeper layers of what that

[00:04:57] sentence was intended to mean. So we're like, oh yeah, let's just move fast, left, right,

[00:05:00] and center. And like, we'll just bounce off like a ping pong ball in a pinball machine, essentially.

[00:05:05] And it's one of those moments where you're kind of like, okay, yes, but let's think about the full

[00:05:10] circular ecosystem here that if you have investors or vendors solution providers, who are kind of like

[00:05:16] pinballs in a pinball machine bouncing off, think about the end users, think about the HR team that

[00:05:21] are kind of chasing after those trends, not exactly sure where things land, not having the time to

[00:05:26] necessarily investigate. And then on top of that, think about the ultimate workforce that's on the

[00:05:31] receiving end of that technology. I think the impact is exponentially larger in that case. And I'm

[00:05:38] not sure if every HR tech designer out there thinks about that when they're launching new solutions.

[00:05:44] Yeah, no, I would absolutely agree. This is a huge challenge. I mean, we've seen it,

[00:05:48] you know, when I go to some of these events and walk around, you know, some of the booths,

[00:05:53] I'm always going in there with a very healthy dose of skepticism. And do the people manning,

[00:06:00] you know, the booth even know how to properly answer some of those questions?

[00:06:06] They don't.

[00:06:07] They don't.

[00:06:07] They absolutely don't. So here's the most iconic conversation I have had. I will not name the

[00:06:13] vendor necessarily, but it was basically demo, demo, demo, fun things, really cool UI, everything else.

[00:06:19] And then I asked a question. I'm like, okay, so what's powering this in the back? Is it pure LLM?

[00:06:24] Do you have a vector database or do you have a graph database? And they were like, yes, all of it.

[00:06:30] And that was my moment where I'm like, okay, I'm going to politely walk away now because that was

[00:06:35] a or question. It was not a and kind of question. So it was kind of that moment. I'm like, as a sort

[00:06:42] of person representing the company, and this person actually carried a title for the company as well.

[00:06:46] So, you know, they're not like a hired marketing kind of help. It was that moment where you're

[00:06:50] just like, do you even know what you've built? And if you don't know what you feel, how do I as a buyer

[00:06:57] trust that I can get what I need to do done? And you're actually there to help me solve my problem.

[00:07:02] And you're not just selling the buzzwords in the market.

[00:07:05] Yeah. I brought this up. Let's see. I was at HR tech 2022 and 2024. So before HR tech 2023,

[00:07:14] I knew I wasn't going, but I put out a post and I was like, you know, AI everywhere. You're going to

[00:07:19] see it in 500 booths. Like it's time to start asking better questions just as you did. And it's not like

[00:07:27] I gave you, here's your cheat sheet, you know, ask these, you know, 10 specific questions, but I did

[00:07:31] have some in mind because as when I was at talent tech labs, we had, we helped our clients with the RFP

[00:07:37] processes and stuff like that. And so it was like, you know, these are some of the questions that you

[00:07:40] should ask. And then we had to basically insert, you know, we augmented that with a set of questions

[00:07:46] to start asking about the models and, you know, can we trust the data or, you know, basically are

[00:07:51] you being responsible by design before I was actually using that, that phrase, but, but anyway,

[00:07:56] I was giving people advice to start asking better questions and I, I missed that show. But then this

[00:08:01] year I went to quite a few events and it seemed like it's, it was like deja vu all over again.

[00:08:07] Like nobody listened to me and the people are still, we're still letting people get away

[00:08:12] with these things and you, you can't explain it. You can't just put out a statement that says

[00:08:16] you're, you support, you know, ethical AI and human centricity and all that stuff without putting

[00:08:22] your money where your mouth is and, and actually going through the steps and showing that, you know,

[00:08:27] you've either been voluntarily, you know, independently, you know, audited or whatever. But I just felt

[00:08:33] like in the, going forward, this is what it's going to mean to be a trusted partner,

[00:08:36] right? You've taken those steps before you just, you know, basically, you know, threw it over the

[00:08:41] fence and implemented it or sold it and implemented it at your, your clients. Like this is important.

[00:08:49] And legislation is always going to be trailing, but responsibly as a lot more than just complying

[00:08:55] with a piece of legislation. For sure. And I think the thing about responsibly, I was telling somebody

[00:09:00] this last week or maybe two weeks ago, I was like, you know, I'm even having whiplash with AI. And I

[00:09:06] literally spend all of my waking hours in the HR tech industry, looking at technology, looking at what's

[00:09:12] the next iteration of HR and work technology. The fact that I'm the, when I forecasted something to

[00:09:17] happen in about like the next 12, 16 months, and in fact, it's happened in the last sort of six to

[00:09:22] eight months. And I'm sitting here going like, oh my gosh, we are moving so fast right now. Part of me

[00:09:28] is also thinking, what does that mean for your typical HR buyer, right? Like the person who's

[00:09:34] working 10, 12 hours a day, who's maybe buying technology on the side of their desk, trying to

[00:09:39] coordinate procurement, leverage their political capital and ultimately get to the right answer.

[00:09:43] What does the world look like for them right now, besides looking at the shiniest brand with the

[00:09:49] most marketing dollars and kind of like, are you really buying the solution or are you just buying

[00:09:54] the sales pitch of the solution? Yeah, I do think they should be put through obviously the same

[00:09:59] rigor as, as anything else. And you're right. I mean, I've definitely seen, well, that was, I guess

[00:10:05] part of that is just how the VC money was coming in and them having shiny object syndrome as well. I mean,

[00:10:11] certainly contemplated pivoting a little bit, my advisory work to be like, look, VC is like, this is part of

[00:10:19] your risk mitigation strategy. This is part of your due diligence. Like, do you just really like,

[00:10:23] oh, agentic workflows? Oh, these guys have a cool website. Like take my money, right? Come on.

[00:10:30] Right. Like you're part of the problem. I think there are a few smart ones out there who have already

[00:10:35] identified and recognized that, but I think there is a lot of sales pitch happening, right? Because

[00:10:41] here's the thing. I dabble a little bit with investing in HR tech as well. As an investor,

[00:10:47] you look for something entirely different than as a solution provider or as a founder.

[00:10:52] And as a buyer, that means something that's entirely different to you. So my struggle, and quite frankly,

[00:10:58] the ultimate goal for what's that I think that out loud is how do you bring all of these perspectives

[00:11:03] together so that you're delivering products that are not only generating positive returns on the

[00:11:08] upstream investment side, but also in terms of the downstream usability side, you're not selling

[00:11:13] a dream. You're actually selling a reality. And it's a reality that's actually going to help your

[00:11:18] end users instead of burning through their political capital, running like six months long

[00:11:23] implementations, and then delivering something that's maybe 25% of what they bought or they thought

[00:11:28] they bought. Absolutely. You know, I think the, as I think about some of the events I went to

[00:11:33] this year, I was at recfest. I was at HR tech. I was at unleash in Las Vegas. And that was a

[00:11:41] people analytics world where we ran into each other in the city. And I don't know if it's just

[00:11:47] like the intimate nature of something like the people analytics world where people are just having

[00:11:53] more seemingly more like real, you know, conversations, but I thought it was a good event.

[00:11:58] I was only able to attend one of the two days in New York city, but it was just a good learning

[00:12:05] experience for me. I mean, I'm not, I don't have a people analytics background. In fact, I don't have an

[00:12:10] HR background. So I'm sort of Monday morning quarterbacking this whole thing. But my background

[00:12:16] is in enterprise transformation, but it's really interesting to see. And on some level frustrating

[00:12:22] to see where he's incredibly smart people. They have the data, they have the insights they have,

[00:12:30] they should have the, the ear of leadership, senior leadership. And yet people analytics teams,

[00:12:38] talent intelligence teams, it's like, still seems like a luxury, especially like mid market and below.

[00:12:45] And I just, I guess, curious to get your impressions of what you heard there. I know you participated in

[00:12:50] one of the debate panels to kick off day two about whether AI and recruiting is better bad.

[00:12:56] Let me bring out my little soapbox so I can stand on that for a moment. I have so many things to say

[00:13:02] about people analytics as an industry. I think starting with the events, so absolutely love

[00:13:07] people analytics world. And you're right in that. I think the intimate setting does bring out

[00:13:11] much better conversations and allowing sort of for that sort of closer transition time instead of

[00:13:16] running between expo halls and sessions and expo halls and sessions and so on and so forth.

[00:13:21] Yeah.

[00:13:22] One thing I have always, that's always challenged me with these types of events though,

[00:13:28] especially like when you're speaking at those types of events and in listening to those types of events,

[00:13:32] is that everybody highlights the 1% of the iceberg. You hear the success cases, you hear how amazing

[00:13:39] organizations have done and delivered. And I've created and have delivered on those promises. But I think

[00:13:45] when you pull all of the speakers aside and we kind of go backstage a little bit, all of us acknowledge

[00:13:50] the fact that we didn't share the 99%. We didn't share the 99% of quagmire,

[00:13:56] of politicking, of hallway questions and those types of conversations that needed to happen.

[00:14:05] And part of me having heard sort of where the success cases are and what sort of everyday

[00:14:10] practitioners are struggling with does question, is it because we haven't taught people how to

[00:14:15] navigate corporate politics? Is it because we haven't taught analytics practitioners who are trained

[00:14:21] to leverage the data, trust the data, because the data means and speaks everything that when you share

[00:14:27] the data, you need to watch the room. You need to understand their context. You need to understand

[00:14:32] your stakeholders before you just let the data do its thing. So I think there's a little bit of that

[00:14:38] in there for me. And I think the other part of it too is, yes, it's a luxury, but it's almost a

[00:14:45] needed luxury. And I have always had this question, and this is my sort of hypothesis that's neither

[00:14:51] been validated or invalidated, if you will, is when you have a more junior level people analytics

[00:14:57] practitioner, because very rarely do you see VPs of people analytics in organizations,

[00:15:03] aren't we covering the data or the true potential of the data through different layers of org hierarchy

[00:15:09] so that by the time it actually makes it into the C-suite conversation,

[00:15:13] whether we like it or not, out of personal interest, out of technicality, and something

[00:15:17] else in between, the richness of the data, of the insights never carry itself to the C-suite. So you

[00:15:24] always have the parties who are wanting the data and the analysis, like the business executives,

[00:15:29] and you have the parties who are trying to deliver, but not quite understanding why it's not

[00:15:34] landing the way it needed to in the business. So a couple things come to mind as you describe

[00:15:42] that. One is that maybe, you know, part of this is about better storytelling with the data and making

[00:15:50] it relatable, understanding the breadth of the problem or the challenge, and that, you know, making

[00:15:58] sure that that hits home. Because just from some of the strategic planning work that I did at

[00:16:04] NBC Universal, that was really part of it, telling that story. And then if you're talking about,

[00:16:10] if you're applying numbers to that, you're talking about, you know, cost avoidance, cost savings,

[00:16:15] you know, things like that. Also, like sort of taking the proposed win, you know, based on whatever

[00:16:21] your projections are. And then also saying, now imagine, continuing the story, right? Say,

[00:16:27] right, Nick, now imagine if we reinvested 25% of that in upskilling, or, you know, this other

[00:16:34] solution that's going to get us from, you know, robotic process automation to intelligent automation,

[00:16:40] or it's going to get us from, you know, these pilot projects in this particular domain.

[00:16:45] And now we can start to, you know, propagate that and drive adoption in other areas and things like

[00:16:52] that. So I do think storytelling is a factor. The other thought I had was around, I guess,

[00:17:00] going back probably 15 years or so, like the advent of social media and collaboration platforms,

[00:17:07] you know, sort of the consumerization of that inside an enterprise that IBM used to start working

[00:17:13] out loud, right? Like people were blogging, people were getting into forums, and they were talking

[00:17:17] about some of these things. And the whole point was to sort of level or flatten the organization so

[00:17:23] that, you know, you could be chatting back and forth with an SVP on a particular topic, and they

[00:17:30] would have the visibility to how conversations were really happening. And then the sort of social

[00:17:37] analytics tools to do that, going back to your point earlier, your comment earlier around the

[00:17:42] graph databases, I mean, IBM research gave us, you know, plenty of those as IBMers. And so it just

[00:17:49] seems like at 2024, now doing planning for 2025 and beyond, like, it's just like, I don't know,

[00:17:57] I guess I get anxious to see, we're moving fast on the AI development front. But obviously,

[00:18:05] the harder part is the behavior change and the actual transformation.

[00:18:09] Yeah.

[00:18:12] You know what you should know? You should know the You Should Know podcast. That's what you should know.

[00:18:19] Because then you'd be in the know on all things that are timely and topical. Subscribe to the You

[00:18:24] Should Know podcast. Thanks.

[00:18:27] And I think it's almost like an industry and a culture transformation that we're asking for,

[00:18:31] right? So what surprises me the most today is once in a way, I kind of peruse people analytics,

[00:18:37] jobs postings just to see what's in the market, what's everyone asking for. The rise of what I call

[00:18:43] the Excel spreadsheet expert requirement and like must be an expert in Google Sheets. I am not joking.

[00:18:50] I read this last week, must be an expert in Google Sheets. It floors me a little bit because

[00:18:56] decades later, we still, for some reason, think that reporting equates to analytics. And the debate

[00:19:02] between a VLOOKUP and HLOOKUP and an XLOOKUP is still thriving. And it's a necessary step. I think

[00:19:09] it's a necessary step in every single organization's journey towards analytics. But the fact that

[00:19:14] some of what I would consider larger enterprises are still stuck in that stage really makes me question,

[00:19:22] how is HR in that particular organization having conversations around ROI, around value delivery of its

[00:19:30] workforce? Or have we just kind of given up and went like, yep, cut my budget, do whatever you want with me.

[00:19:35] I'll just sit back because it is what it is at this point.

[00:19:38] Yeah, I've seen, you know, I poke around periodically as well, especially on the like responsible AI, AI

[00:19:45] governance kind of space because I'm really curious to see like the overall sort of adoption. Am I seeing

[00:19:54] increasingly more organizations are looking for people to fill those roles? If they're not, is it

[00:20:00] because they are just complete laggards or have they actually said, yeah, I think this is important,

[00:20:08] but, you know, Joe and in legal and Sue in, you know, HR compliance, you know, they, they got this,

[00:20:15] right. So I do pay attention to that too. And I, even these leadership positions overseeing some of these

[00:20:22] potential groups, they're still asking the same sort of, I wouldn't say 20th century skills. I mean,

[00:20:30] Google sheets is not a 20th century skill, but the point is when you think about skills in general,

[00:20:37] and your skills gaps and your skills needs, you have to incorporate some of that trajectory of AI

[00:20:46] capabilities. I mean, it may not be great in certain areas now, and certainly you and I can pick apart

[00:20:53] all kinds of tools that are available today. But if you look at the pace at which these things are

[00:20:59] evolving, I mean, really, I mean, these are, these are the skills that are either must haves in your

[00:21:07] job description. I think you should maybe rethink that.

[00:21:11] Yeah. And honestly, on the topic of skills, that's actually what scares me the most about HR in the age

[00:21:18] of AI. Quite frankly, it's not whether HR is using the appropriate technology, responsible AI,

[00:21:23] and the whole nine yards. I think, oddly enough, the HR tech industry, as it has done in the last

[00:21:28] decade, two decades, will eventually guide the HR department to the right path in that regard.

[00:21:33] What concerns me is in the age of AI, when you have the people parts of your organization,

[00:21:39] not understanding the skill sets, the possible depletion of skill sets that takes to run your

[00:21:47] organization. That's a very scary concept, because imagine if you were a HR business partner,

[00:21:54] a recruiter sourcer, who got handed a job rec that says, I need to hire a prompt engineer.

[00:22:00] Where do you even start? Where do you even start with a draft job description? Where do you even start

[00:22:04] with? Okay, what are they prompting? What kind of language model are they using? Like, what's the

[00:22:08] backend architecture? How do you even start asking the business and having those types of conversations

[00:22:14] with the business? And sure, one job description isn't much. But when your business starts to advance

[00:22:19] faster than HR departments do, you're going to have multiple of these conversations. And eventually,

[00:22:25] I do think it will become a credibility issue from an HR function standpoint of, are you mature enough

[00:22:32] to support the business? Or is this one of those times where the business will have to figure out how

[00:22:36] to support themselves again? That should be a serious warning to HR teams, right? To get their act

[00:22:43] together, because they're already fighting against, you know, reputational issues, right? Like,

[00:22:51] they're not, are they really helping to drive change? Are they really my talent advisors? Are they really

[00:22:56] helping me with, you know, strategic workforce planning and some of these longer term, you know,

[00:23:02] existential things? Or are they just, you know, firefighting and protecting the organization from

[00:23:08] all kinds of risk? I mean, you've got to, you've got to do it all, frankly, and you've got to stay

[00:23:14] on top of this, you can't relinquish control. This is not a technology, I mean, yes, it's AI, but

[00:23:19] it's not exclusively a technology challenge. I mean, HR knows this, right? So it's just a matter of

[00:23:26] how are we going to position ourselves? And as you pointed out before, like, this concept of,

[00:23:33] you know, social capital, political capital within the organization, how can you maintain and grow

[00:23:39] your influence? Because, because if you don't sort of embrace this opportunity, you're really going to

[00:23:45] be behind the eight ball, as they say. Yeah, exactly. And I think it's even like a lot of times I have

[00:23:51] conversations with HR professionals who are like, oh, you know, my job is compliance, my job is risk

[00:23:55] mitigation. Like, I don't need to worry about advancing technology or like, what is AI doing

[00:24:00] to the workforce? And I'm sitting here going like, okay, but if you don't understand what's

[00:24:04] happening, let's talk risk and compliance for a bit. The EU AI act is already in play. We know,

[00:24:09] given GDPR trajectories and the hoopla around that, it's a matter of time before it rolls out

[00:24:14] in North America in the, what I call the 50 flavors of the United States at a certain point.

[00:24:20] Yep.

[00:24:21] It's a matter of time that happens. Without understanding technology, how are you supposed

[00:24:25] to understand the regulation that's behind it? Number one. And on the whole risk mitigation side

[00:24:31] of it, if your employees are leveraging AI technology as part of their personal tool stack,

[00:24:36] but on your company equipment, how are you supposed to write policies around it to kind of help protect

[00:24:42] the company, keep its data private, keep your company data from training the sort of larger language

[00:24:47] models out there and looking at that sort of sandbox periphery? Because sure, IT can take some of the

[00:24:54] technical heavy lifting, but let's be realistic when it comes to policy drafting, enforcing the

[00:25:00] policies and kind of monitoring. A lot of times that ends up on HR's desk. And I think it's also a

[00:25:05] question of, even if we break HR down to its most foundational elements of compliance and risk

[00:25:11] mitigation, there is still a bit of, I think, a learning trajectory and maturity curve that needs

[00:25:16] to happen, especially when you start folding in the conversation of ethical and responsible AI amidst all

[00:25:22] of that.

[00:25:23] Yeah. No, it's definitely a lot. We are all responsible, right? And that's why we need to be

[00:25:28] responsible by design. That's why we need, I'd still contend it's inevitable that we'll have

[00:25:33] like sort of onboarding and regular compliance training around, you know, ethical AI, AI governance

[00:25:39] and things like that, just like we do with privacy and cybersecurity and harassment. I just think everyone

[00:25:45] needs to know those basics, whatever your familiarity with, your comfort level is with

[00:25:51] the technology itself. You've got to know when to trust the output and how much to rely on that when

[00:26:00] you use it as part of a decision-making process, especially in talent and HR where we're talking

[00:26:05] about people's livelihoods and, you know, there's human beings behind, you know, those rows in your

[00:26:10] TTS and your other systems. Exactly.

[00:26:13] Oh, I wanted to ask you about person's Galileo.

[00:26:17] Yes.

[00:26:18] For people that didn't watch the video, I encourage everyone to go to YouTube and watch

[00:26:21] this video. It's not very long, but just, you know, your, your quick impressions. And then I was

[00:26:28] curious if you have also tried like Atlas Copilot or any other, you know, similar tools that are gen AI,

[00:26:35] but within a specific domain, you know, focused on, on talent, I guess.

[00:26:40] Yeah, for sure. So I dabble in AI tools quite a bit because ever since I kind of got into the

[00:26:46] solution provider side of the house, I realized that I have now become a firm believer that if I

[00:26:51] can't touch it, I can't see it. Your marketing language does not exist to me. I feel like I've

[00:26:56] read and seen too many things where I'm like, I don't think that's what you meant to say. Or like,

[00:27:00] I don't think that's what your product actually does. So that's how the whole Galileo thing

[00:27:04] started. I came across Galileo when it was actually launched in April, May this year at

[00:27:10] the enterprise level, essentially. I thought it was really cool, but at the same time, I'm like,

[00:27:14] oh, that's another, at least five figure enterprise contract. Like nobody else can take advantage of

[00:27:19] it. It's kind of one of those things where you're like, yep. And there's a Delta between the larger

[00:27:23] companies and the smaller ones once again. So when they announced that Galileo was coming out with

[00:27:28] individual licensing, I think about three, four weeks ago at this point, I was like, oh,

[00:27:32] this is really interesting. Let me sweat my credit card, pay for it and dabble with it. Right. Because

[00:27:36] everyone gets to wipe a credit card these days. So I'm like, let me grab a license and see what

[00:27:40] this thing does. Yep. I think it's genuinely cool. And it's genuinely cool, not because it's AI. I think

[00:27:46] the AI to me is like that 5% cherry on top. It's the fact that Burson's research is now available

[00:27:52] for public consumption in a very targeted format. And I think when you start looking at sort of the macro

[00:27:59] economic factors in our industry right now, it's almost a perfect storm and the perfect timing for

[00:28:06] that solution. Because more than ever, you're seeing VPs of HR, CHROs stepping out of their roles.

[00:28:11] There is a rise of the fractional that's happening in the market right now. Simultaneously, you're

[00:28:16] seeing internal to organizations where the average tenure it takes to get into a more senior role

[00:28:21] isn't that long anymore, which I think is a phenomenal thing. But it also means that there isn't

[00:28:26] a full suite of rotation, I think, across the HR function before someone is elevated into senior

[00:28:31] leadership position. So in a way, what Galileo does is it allows you to kind of access almost two decades,

[00:28:38] if not more than that, worth of HR research at your fingertip when you need it in digestible language

[00:28:48] that does not feel like you're reading an essay. And on top of that, it's with sort of like all the

[00:28:54] resources and all the supplemental documents attached in the background. So you can easily

[00:29:00] click into the next layer down if you need to. I think it's really playing into the fact that

[00:29:05] our attention spans are getting a lot shorter. We're just not reading 20 page research papers anymore

[00:29:10] in this day and age. And we just need a instant sort of gratification that when we have a question,

[00:29:16] we need an answer immediately. And I think we're going to see more and more of that, quite honestly,

[00:29:21] in the work tech market, where the sort of consumer market carryover or patterns, if you will,

[00:29:28] in terms of attention spend, as well as the desire for instant gratification and possibly infotainment,

[00:29:33] as we now call it, is going to come into the work tech market.

[00:29:38] No, I think it's certainly conceptually, it's amazing because, you know, Josh and his team have been around

[00:29:44] for quite a while. And to have all of that, I'm assuming there's no, are there any particular

[00:29:51] limitations in terms of like the types of content or anything that they've researched is fair game?

[00:29:58] So I have yet to come across something that is a sort of general interest topic that hasn't been able

[00:30:06] to give me an answer in Galileo. And I've been playing around it for about a month at this point.

[00:30:10] There are some like more niche questions that you'll ask and I'll be like, oh, I'm sorry,

[00:30:14] I'm not trained to answer those types of questions. But I would say that as a new CHRO or VP of HR or even

[00:30:22] HR business partner, it will minimally get you 75% of the way there. And then like once you know the

[00:30:28] language and like the thinking pattern behind it, I think it's pretty easy to fill the rest of the 20%

[00:30:33] gap either with Google perplexity or pick your weapon of choice essentially to augment that level

[00:30:40] of research. Have you ever been to a webinar where the topic was great, but there wasn't enough time

[00:30:46] to ask questions or have a dialogue to learn more? Well, welcome to HR and Payroll 2.0, the podcast

[00:30:51] where those post-webinar questions become episodes. We feature HR practitioners, leaders, and founders

[00:30:56] of HR, payroll, and workplace innovation and transformation sharing their insights and lessons learned from

[00:31:02] the trenches. We dig in to share the knowledge and tips that can help modern HR and payroll leaders

[00:31:06] navigate the challenges and opportunities ahead. So join us for highly authentic, unscripted

[00:31:11] conversations and let's learn together. Yeah, no, that's pretty impressive. Well, first of all,

[00:31:17] I've seen a couple of different examples of where people have built Gen.AI tools for market research

[00:31:25] more broadly. And I think that's an amazing use case. I think that if you have the attribution

[00:31:32] and understand where the sources are all coming from, I think that's really valuable. And you can only

[00:31:39] imagine what you would pay to have somebody go off and do that level of research, particularly with

[00:31:44] the accolades that Josh has earned over the years. Well, and the thing is, it also does part of your

[00:31:50] job for you, right? So they have these little templates that they've set up in there. And I think

[00:31:54] the one I've been playing around with, it's sort of like global PTO policy template or something like

[00:31:59] that. And essentially, it'll write you a PTO policy, like you enter the country, it will kind of get you

[00:32:04] like pretty much again, 75% of the way there in terms of what the government requirements are

[00:32:09] pertaining to a specific state, a specific province and a specific country from a PTO standpoint,

[00:32:14] from a leave of absence standpoint. And I thought that was pretty cool. It's not that you can't get

[00:32:19] that stuff off of Google, but just that when you keep it in the same place, it's almost like you can be

[00:32:24] like, okay, I'm going to go really deep and like really strategic. And now I got to go executional

[00:32:29] on the policies and it's just all there. That's interesting. Okay. Cause I thought maybe the main,

[00:32:34] not the main difference, but one of the differences between like, uh, and I haven't played with Atlas

[00:32:39] co-pilot to compare it necessarily, but I mean, just for the audience sake, Atlas co-pilot was released

[00:32:44] by, um, HR leaders, Chris, Chris Rainey and team Matt Burns, I think was involved, but I thought

[00:32:50] Burson's was more about his vendor and market, uh, research. Whereas an Atlas co-pilot would be

[00:32:58] about like what you were describing, like policy based stuff, uh, like this Uber HR virtual assistant

[00:33:06] kind of thing, but it sounds like Galileo does.

[00:33:10] Galileo's got a bit of both. Um, I'm actually going to go see if I can get a license to Atlas

[00:33:14] co-pilot now, and we might just do a compare and contrast YouTube video at this stage to see who does

[00:33:20] leave of absence policies better. Um, I haven't dabbled in Atlas co-pilot yet, but I think so Galileo,

[00:33:26] if I were to kind of assign a percentage breakdown, it's probably about like 70, 75 theory. So what you

[00:33:33] would typically expect of Josh Burson company, their research, that type of thing. And then 25%

[00:33:39] practice where you can actually search up benchmarks, um, from different industries through this year.

[00:33:44] And I think there's a couple other skills, taxonomy skills, mapping stuff, I believe from

[00:33:49] light cast that's in there as well. So Galileo is really combining, if you will, all of the

[00:33:54] industry's knowledge, um, and putting a practitioner's lens on it and then distributing it through that

[00:34:00] channel. Okay. So that's, that's encouraging. Cause I, one of my concerns and nothing, of course,

[00:34:08] nothing against Burson. I mean, it does amazing research. It was more, it was more just, is there

[00:34:14] inherent, you know, bias in the data that's now being summarized by the same people that had the

[00:34:23] same team. And now you have this one lens, right? But it sounds like that's not actually the case.

[00:34:30] It's, it's incorporating other data to supplement that.

[00:34:33] So there are certain papers that the Josh Burson company publishes that kind of make me like,

[00:34:38] I see where you're coming from, but have you thought about?

[00:34:41] So even in those types of topic areas, when you kind of look at the way, um, Galileo lists out the

[00:34:47] answers without going deep in the research paper, there's a level of neutrality in how those answers

[00:34:53] are shaped and from a natural language format that upon reading the paper, you might not necessarily

[00:35:00] get. So I think there's also a level of sort of tweak that was done in the backend as well. So I,

[00:35:06] I like it to be honest, like I think it would really benefit the fractional, um, to see Charles out

[00:35:11] there. Um, and really like any HR general generalist or any HR business partner, cause it's like your

[00:35:17] brain's going to feel like it'll melt after 30 minutes reading that stuff, but it is also going

[00:35:22] to make you a lot smarter after 30 minutes of reading that stuff.

[00:35:25] No, that's, that sounds pretty compelling. The other thing I would just give them, you know,

[00:35:30] at least one bonus point for is to actually acknowledge that it has its limit, its own limitations,

[00:35:35] right. I, it drives me crazy when you ask some of these Gen AI tools, a question and it doesn't know,

[00:35:43] but it gives you an answer anyway and makes it sound like it knew. And if you don't, you don't know enough

[00:35:50] about the subject matter, you're more likely to trust its response. And I feel like perplexity has

[00:35:56] gotten worse with, in that regard. Like, I don't understand what it's doing, but it's apologizing to

[00:36:03] me constantly because I keep calling it out. No, I shouldn't have said that. Oh no, that's right. I

[00:36:07] did. I did just completely make up that company that you were looking for that you couldn't find.

[00:36:13] I just completely made up what they did. Like what? Oh my God. Just say that you don't know,

[00:36:19] just stop the nonsense. Yeah, definitely. I think for me, perplexity, if anything is just a quicker

[00:36:28] redirecting tool at this point than a Google search. I wouldn't say like, I trust a hundred percent

[00:36:32] the answer that comes out of it or even 80%. Cause sometimes when you kind of start going

[00:36:36] through the financial figures and stuff like that, it'll give you data. And then you're kind

[00:36:39] of sitting there going like, okay, but like, are you actually interpreting the core source of that

[00:36:43] data as it was intended to be interpreted or just taking that one sentence and kind of going like,

[00:36:48] okay, yep, this is what I'm going to serve this for you. The other thing that comes up lately,

[00:36:53] not in terms of hallucination, I wouldn't, some might call it a hallucination, but like these

[00:37:00] summarization tools where it may not necessarily say something that's not, that's false, but it,

[00:37:08] it draws, it could draw conclusions that aren't grounded, or it could take a little bit of creative,

[00:37:17] you know, liberties just to sort of like, if you were to say, if you didn't like the response and

[00:37:22] you say, expand on this, right. All of a sudden it starts getting a little too creative. Right.

[00:37:30] And it's like, well, where'd you, where'd you come up with that? I mean, it's happened to me just

[00:37:35] trying to, if I ask one of my go-to tools to do like a summarization of a podcast based on the

[00:37:43] transcript or the show notes or whatever. And it's like, not only at no point did we talk about

[00:37:49] that subject that you just wrote a paragraph on, but you overlooked and didn't include

[00:37:57] some of the best parts of the discussion. So that kind of stuff scares me, not necessarily. I mean,

[00:38:04] yes, it, it could impact you doing, you know, market research, but imagine if that was the case in

[00:38:12] an interview transcription tool or something like that. So I don't know, it just, it's concerning,

[00:38:19] right? It's concerning that there was an article the other day in the Associated Press,

[00:38:23] Hiltke, Shulman coauthored a piece in the Associated Press around his health, one of these

[00:38:29] healthcare transcribing tools, completely made up stuff, right? In a, in a patient,

[00:38:36] in a doctor's summary of a patient interaction that is unacceptable. So I just think it's still early

[00:38:44] days for a lot of these things. And, you know, if we're going to advance so quickly, can we not,

[00:38:51] can we not break the patient doctor, you know, trust that would be a good thing if we didn't,

[00:38:57] if we didn't do that.

[00:38:58] Yeah, for sure. I think it's fascinating because as you were saying that I just realized something

[00:39:02] and this is a completely personal preference, but I realized I don't use AI to summarize things.

[00:39:09] I make AI expand on things. So sometimes I'll just have an idea because apparently I've been told that

[00:39:15] I talk and think in bullet points. And as such, it's really hard for me to create paragraphs or like

[00:39:20] event descriptions and stuff like that. So I'll dump a bunch of bullet points and I'll get AI to

[00:39:24] put together a summary or a synopsis of sorts. And it does a really good job in expanding those ideas.

[00:39:31] But I can see where the challenge is when you're trying to get it to summarize concepts and ideas,

[00:39:35] because depending on the language model that is trained after and all of that fun stuff,

[00:39:40] I think the complexity of the human thought process has yet to be fully translated into AI format.

[00:39:50] So it's almost like it's good for repetition sake, for expansion sake, for like ideation sake,

[00:39:55] but not necessarily to like take over for part of the work. So like health record transcribing,

[00:40:01] for example, or like summarizing podcast and things along those lines.

[00:40:05] Yeah, no, that's fair. I do play around with it a lot of different tools. I see a lot of

[00:40:14] consistency and not in a necessarily good way across some of these tools. They all, you know,

[00:40:20] operate, you know, similarly, and that's across different LLMs as well. I mean, I know people have

[00:40:27] been doing like you were doing with your video, like analyzing different tools and comparing results.

[00:40:34] I'm curious if you've played around with any of the ones. There's some new tools cropping up that claim to

[00:40:41] basically you put in your prompt once and it'll go and basically execute it across multiple models.

[00:40:50] Have you played with any of those?

[00:40:52] I've dabbled with this. So I call it daisy chain AI for lack of a better description. I think there's multiple tools

[00:40:59] that do that. And I think how every tech platform daisy chains the backend together significantly

[00:41:06] impacts the output that you get. So on the good ones, for example, I can basically prompt like draft

[00:41:13] me a presentation. Here's what I'm looking for. Here's how long I have, et cetera, et cetera.

[00:41:18] And I will get a presentation that's not necessarily camera quality, but something that you can dump

[00:41:23] into Canva and have Canva make look nicer. And then there are others where I think it was last

[00:41:30] week, one of the images came up with six thumbs and I'm kind of sitting there going like,

[00:41:36] not quite sure how we got the six thumbs here, but that is so not what I was looking for.

[00:41:41] Yeah. That's not good. Yeah. I mean, it's sometimes, you know, you just have to take a step back.

[00:41:47] You know, we're in this sort of echo chamber where everyone's like following every little

[00:41:51] point release in every development so closely. We've got to take a step back, maybe talk to,

[00:41:58] you know, non-techies, talk to people in other domains and really understand,

[00:42:01] you know, what, what they're seeing. I mean, the healthcare one came up sort of by accident

[00:42:07] across, across my path because I was doing research, but, you know, I talked to people in education.

[00:42:12] I'm going to be talking soon to my local, you know, school district and folks running curriculum

[00:42:18] and technology or whatever, because I feel like, you know, the sooner we can get people,

[00:42:22] you know, comfortable, then we can work together to carve out the right policies. But just sticking

[00:42:29] your head in the sand is not a policy and it's, it's not helping our kids be prepared for what's

[00:42:36] next. I mean, we're going to start to see these job descriptions instead of saying, you know,

[00:42:41] you need to be an Excel jockey and, you know, master. When I was growing up, it was, yeah, it was,

[00:42:48] it was, you know, do you know, Microsoft office? Do you know, whatever, how to use a computer?

[00:42:54] Right. I mean, I had a computer, I mean, I'm, I'm old now, but I mean, I was given my first

[00:43:00] computers in school and when I was in middle school when I was 11, 12 years old. So, yeah,

[00:43:06] I mean, some ways, you know, using these AI tools is similar to that. So, you know, my daughter's 16

[00:43:14] and she doesn't get to use it at all related to school. And so I want to change that, but, you know,

[00:43:22] going out into these other domains, talking to people who are not tech savvy and who are just

[00:43:28] experiencing frustration. I mean, we see the adoption rates, adoption rates are higher than,

[00:43:33] than computers. It's like social media adoption or maybe even higher. Right. So the train has long

[00:43:41] left the station. Right. And so we're already sort of backtracking and trying to make sure people,

[00:43:48] imagine if just, you just cars came out and you didn't need a license and everyone was just

[00:43:52] driving around in the car. Yeah. I mean, I think that's a really good point because one thing I do

[00:43:57] appreciate the quick rise of AI for is bringing back the concept of what I call play, right? Because

[00:44:04] sometimes when we enter the working world, and I think it's true for a few of the generations that

[00:44:09] are currently working, it's like, you're told that this is a model. This is a mode. You have to do

[00:44:13] X, Y, Z step. I mean, we create a career path for a reason. You have to do X, Y, Z step to get to

[00:44:18] the next stage and then X, Y, Z step to get to the next stage, so on and so forth. But the thing about AI

[00:44:23] and the amount of data it takes to power AI, it's like what used to be linear is a bunch of zigzags

[00:44:29] and could entirely be circular at that point. And it's kind of encouraging everyone to go back to

[00:44:34] the concept of like play around with it, experiment with it. And it's almost that sort of lost sense

[00:44:40] of curiosity, I think, that we need to bring back to the workplace, quite frankly. I imagine that's part

[00:44:45] of your answer when I, if I were to ask you your advice to elevate your AIQ, right? Probably.

[00:44:53] Curiosity is definitely one of the top answers. So yeah, we covered a lot. Lydia, I mean,

[00:45:00] I was going to ask you about some of your favorite tools. So if there's anything you're particularly

[00:45:05] fond of or scared of that we didn't cover already, I'm curious.

[00:45:10] I don't think I'm scared of anything. At least I haven't come across anything that I'm scared of

[00:45:15] yet. Particularly fond of, Weerly is actually not an AI tool, but it's, I'm sure everyone's heard of

[00:45:23] this. If you're on YouTube, these guys sponsor like 10 bajillion YouTube creators, but it's called

[00:45:28] brilliant.org. So it's basically like a courseware training type of thing. It's a gamified learning,

[00:45:33] essentially on a monthly subscription model with like a 14 day free trial or something along those

[00:45:38] lines. But they do a really, really good job of simplifying concepts. Like how do you code with

[00:45:44] Python? What is an LLM? How do you AI work? So on and so forth. And I think for anyone that's

[00:45:50] curious out there, definitely worth maybe even just a one month subscription to kind of check out

[00:45:55] and see what it's all about. Cause short of using the tools, I think it's also nice to kind of

[00:46:00] understand the backend of what's actually happening from a system perspective.

[00:46:05] Got it. No, I'll have to check that out. I'm not familiar with it. I have looked at a bunch of

[00:46:10] different, they all start to look the same, right? I think it's hard. I was just having this

[00:46:14] conversation on LinkedIn just yesterday. We were talking about hacking HR had created like,

[00:46:20] like a 20 question kind of AI readiness assessment. So I commented, like I've gone down this path. You

[00:46:26] can imagine AIQ. We've gone, done some iterations on this, this for sure. And it really depends on how

[00:46:35] you ask the questions and who you're asking the questions of and, and the domain that you need to

[00:46:41] know as well, right? Like an HR business partner doesn't need to know the same things as, you know,

[00:46:46] maybe a product manager or, you know, people who aren't in technology at all and, and your level of

[00:46:52] proficiency in terms of actually working with, with data or, or having, you know, some data,

[00:46:59] you know, background or what aspects of responsible AI are most important for you to understand. I mean,

[00:47:06] that's, it's different depending on your, your role and perhaps your, your seniority or industry.

[00:47:11] But I do think that there's some basics like we were talking about before, like whether it's

[00:47:16] through compliance training or whatever, there are some basics that everyone needs to understand.

[00:47:20] And I do think once people have a good sense of what this can do, you're going to get a lot of

[00:47:28] ideas, right? And then what are you going to do with those ideas? You have a way to sort of capture

[00:47:33] them and potentially realize them by, by investing in them. So there's a lot, I'm optimistic that we'll

[00:47:41] start to go down that path, but we just, we need to, we need to innovate responsibly. It's sort of my

[00:47:46] new little tagline. Yeah, for sure. Lydia, this has been awesome. Thank you so much for, for nerding

[00:47:52] out with me on all these, these topics. We could probably talk for a few more hours, but I'm going

[00:47:57] to let you go. But again, thank you so much for, for sharing your insights with, with my audience and

[00:48:03] chatting with me. Yeah. Thanks for having me. Absolutely. Thanks everyone. That concludes

[00:48:08] another episode of Elevate Your AIQ. Thank you again, Lydia, and we'll see you next time.