Episode 60

Robert Hardman on Reimagining Hiring & Leveraging AI for the Future of Talent Acquisition

TIM: Robert Welcome to the Alooba Objective Hiring Show. Thank you so much for joining us.

ROBERT: No, my pleasure.

TIM: It's really a thrill to have you with us. Last time we were chatting, just a couple of weeks ago, I just had this sense you're going to be very engaging, interesting guests with some unusual views. You're not going to give us the plain vanilla thoughts, and I'm just really excited to chat with you today. And I would love to kick off with an observation that I've made over the past six years, and I want to get your thoughts on it. I've been working at the kind of crossroads of recruitment and data for the best part of six years now, and one thing I've always found quite curious is that I'll observe data leaders These are analytics leaders and data science leaders. I must've met probably a thousand in the past six years, and I discussed with them how they approach hiring, what their philosophy is, and I'd say probably 80 percent of them take a very gut-feel, intuitive-based approach to hiring. despite the fact that in the day jobs, at least seemingly, they're all about data and product data and operations data and sales data and this and that they're preaching, they're up there trying to get data literacy into the organization like they're just, it's like they take off a data hat and put on an intuition hat when it comes to hiring. Have you observed that? Do you agree with that? Can you explain why that might happen? I'd love to get your thoughts.

ROBERT: Yeah, and I think it depends a lot on the level of the role, right? We're talking across a spectrum of roles here about how you would discuss or talk with somebody if you're looking for a senior VP. senior engineer, senior architect, upwards is very different from okay, I'm body shopping; I want to get some juniors, etc. in When it comes to those senior roles, yes, you do see a lot of it is on gut feeling because, first of all, they've never collected the data, so they're quite data poor in the HR area, and these senior roles, a lot of it is, Will I get on with the person when I'm interviewing? It's very much that that is just as important. You're presuming they're coming with the skills, and you generally distill that fairly quickly within a conversation. It's not as if you're asking for specific skills that you might if I'm hiring a machine learning engineer or a data engineer, where you expect them to come with basic skills and that So it depends; it really does. I will say at the senior levels, gut is maybe too strong a word, but very much that personal interaction in my interviews when I take them for senior roles, and that it's a conversation, and often the person said, When does the interview start? I said, We're finished. I've had it. I've got all the information I need to in a good way, and I've had a 30-minute conversation. I know more about them than any standard for those sorts of roles, and I find that has been great, but you can't do that if you've got 800 people applying for a role. I know that, so you focus. It's not one solution fits all; you go all right For these sorts of roles that are going to be like this, I'm going to be interacting with these people very differently from those roles where you're looking for, I don't want to, you know, a set of skills. When I say lower level, I don't mean that in any derogatory sense. Okay, I need programmers. I need software engineers with those sorts of skills, then I feel data can help you sift through the list a lot better to find key attributes if you have that data set if you can find some organizations. You analyze the statements when somebody leaves HR will collect the reasons why you are leaving, et cetera, et cetera. That information can be very rich, but it takes a little bit of science to be able to turn that into useful information, and not many organizations either have the time or the capability to do that. They take these interviews but never use that information.

TIM: So there's a scarcity of high-quality data to make the decisions in the first place. What about if we solve that problem hypothetically, like if we could click our fingers and have this magical universe where we had the data that was accurate that we actually wanted, would then it make sense to have a more data-driven approach even for these senior roles, or even then do you feel like it's still just going to be down to the gut?

ROBERT: I think it's a mixture of both; for instance, if you're able to click your fingers like you said and you have the data, and that's a lot of attributes, and with modern techniques, it's very similar to an LLM. One looks for a similarity space; you put these vectors into space, and you're able to cluster to see these are one type of skill, these ones, etc. So there is a huge potential that can or uses the same mathematics that large language models use, and no one's really started to use that. There's also a moral issue to that, but in a sense, you break the person down into a set of characteristics, and each person has those characteristics, so you remove a bias, and then in a sense, using the mathematics that is used in large language models or similarity, you can start to find clusters of similar things that lead to certain outcomes. Now that will be fantastic; that will enable Shall we say you should remove a certain bias if you do it correctly to discover talent where you may not have found that talent before? It opens up a whole I suppose what you would call it is a drawer of goodies that are very possible for you to look at. It's a wee way away, but the mathematics, the techniques, and the technology are there; the data isn't there readily yet.

TIM: And because, yeah, you have such a strong background in the underpinnings of these models, I'd like to hear you almost unpack that in a bit more detail, so when you say the techniques and the math are already there but the data isn't, what could it look like in five years if we had the data plus the techniques, and what are those techniques in your view?

ROBERT: I'll give an example of not people but let's say cars, and you're pricing cars at the moment someone will look at a car and then go and look for similar cars online or in adverts. That's pretty close to that. Oh yeah, one of those, or I've sold cars for a hundred years; I know gut feeling. The other technique is to say, All right, all cars are the same. They have an engine, they have wheels, they have several body shapes, they have so many windows, they weigh so much, they have an engine size, and they have 30-40 characteristics. When you store those characteristics, when a new car comes along, you don't try and look for the similar You go, It's got those wheels; it's got that many things, and all of a sudden you can price this car you've never seen before because all cars are made up of pretty much the same bits and pieces, and so in that vector space you can price something you've never seen, so that's a sort of example, so in a sense a person's data in terms of what job they've done and how long they've been at it You capture all those attributes, and every person will have a very similar set, and then you can start to look for patterns in those attributes, and we already do that in large language models. We can do it for pricing cards; we can do it to find the characteristics of people for certain jobs if that makes sense.

TIM: It certainly does, and I get the feeling that the average HR talent person would have almost visceral hatred or rejection of this idea of using data science and almost measuring humans, and they're like, Oh no, I want to take a more human approach. I feel like this is somehow dehumanizing. It's like applying math to people. I feel like they're really going to resist this quite strongly, but I assume this is a lot of how the sports teams have improved their recruitment in the last 15 or 20 years, like they've moved to a very fundamentally data-driven model, I would have thought.

ROBERT: You're absolutely right on both points. Yes, this is already being used in some areas, and yes, people first find it alarming that you can reduce things down to their components, but it's not depersonalizing people; it's categorizing attributes of people, and it's different, and it removes the bias that a human, without fail, will apply to anything, and even if they don't even know it, you see it in the data constantly, and we know about it. So for whatever you might lose, you gain a far lot more. That's all I can say. Now how you sell it, it's difficult; people are difficult. Its power is immense, and the potential is excellent for finding talent and finding the right roles for the right people. the cost of getting people into the wrong role I don't even want to try and estimate it on the back of an envelope, but it probably costs organizations 40 percent of their profit if they got it right. So it's worth the effort getting the right people into the right role and not losing people and losing that time; it costs so much more than businesses realize.

TIM: Yeah, so that's just such a huge upside, and I wonder if it will be like some kind of competitive pressure that eventually forces companies to change. There'll be some early adopters; maybe some tech companies will do this, or maybe they are already doing this. They'll start to win. companies will say, Oh shit, like, we can't compete. They know what they're doing; there'll be some kind of flick of the switch, because I feel like that's maybe one reason in sports that it's taken off is that the results are so obvious. It's literally a leaderboard. You can see whether or not your process sucks immediately. and there's very easily measurable data as well. Maybe it's the combination of those two things we need to unlock: the data and then maybe the speed to result as well. Is that an angle, like in sports teams? You recruit two new players, and they start playing well straight away, then you have the result immediately. Whereas in a business maybe it takes longer—is that part of the problem?

ROBERT: I think they all come into play. Getting the right person is the amount of time and the cost of turnover; it's ridiculous, and if you can get 70 percent of the right people through the door, By these sorts of methods, you're going to be a winner. I think one of the catalysts that will cause this change or what it will become in this history is when there's a paradigm shift, and at the moment there's this change that's being brought upon by analytics and AI. where the skill sets are so new, so there's this gap, and I think everybody realized, and they're all just struggling; they don't even know what to look for and hire, so there's a need now because all of a sudden, to be fair, They don't have a clue what the JDs are or what the skill sets are, so they're going to be looking for okay What do we do because you have people who are used to hiring people for jobs that were created in the 20th century? We're a quarter of the way through the 21st century. The jobs are different; what people do is different, but the processes are still late 20th century, and as we start to move into what I call the intelligence ecosystem, where intelligence decisions are being made, it's very different from what we just had not long ago: business intelligence. No, it's a complete change. That gap is going to force a change that's not going to be optional because those who get on board and get those few people who are capable of, you know, Doing these skills will streak ahead, and if you're still sifting through 4000 CVs and you're not quite sure what you're looking for, it's not really going to work, is it?

TIM: And unpack for us, if you can, a little bit more what this transition is that's happening. What do you see happening right now? What do you see happening in the next couple of years?

ROBERT: Well, at the moment, people are using, you know, and I'm generalizing here, so forgive me for that, but it's to get a story across. The positions that people are hiring you for aren't even that valid for the new ecosystems that people are working in. You've got, Oh, I need a data engineer. I need an oldie CTO or a data architect, and the roles are siloed because that's how they were developed in the end of the 20th century, being in the 21st now. When I need somebody in, we'll call it what I call the inte in the intelligence age. You need people who can sit across three or four different skills. They're not siloed; they need to understand data; they need to understand math; they need to understand AI; and they need to think strategically. Somebody comes to me, and they're a brilliant Python coder—probably no use to me. What were skills of immense importance? Now you would need people who can problem-solve, who can look across the full stack. There is no point in being a data engineer if you don't understand what that data is going to be used for, and at the other end, you've got the AI and the data scientists who are going Oh, I need this, but they need to understand how the data comes in and what you can do, so roles that don't exist yet, like an intelligence architect, are made up. That name is made up, but that is a role that is needed, but it doesn't exist. So many of the roles that are needed today don't even have names, and people are hiring for these old silo positions and wondering why they're not working because they don't understand how the new ecosystems are starting to evolve and develop and the types of schools you need. It is a very new world.

TIM: And if I just put on my economist cap, I say I studied economics, and I'm thinking straight away of division of labor specialization in the industrial revolution. You start to get the pin factory, where you decide we're going to be more efficient if everyone does their own little part rather than one person doing a bit of everything. and but the picture you've just painted is that now we're going to move out of these siloed roles into this broader skill set. What do you think that needs to happen, and why is that better? unquote

ROBERT: Okay, the industrial evolution was making widgets. It was all about math and how many I can make for a unit cost type of thing. The world today is all about taking value out of the data that exists. The data is more valuable than the company often, and the skill sets you need to do that are a broad set. It is not an individual, and people have to retrain throughout their careers. Now you need people who are fluid in their skill sets. They've got the basics, and you can see that they can learn technology changes so quickly now that what used to take three years or four years to go for a cycle is taking 18 months. You have projects that take 18 months to plan, and you look at them, and you realize you'll be out of date before you even finish your project. That is how quickly the technology ecosystems are overturning. It is a completely different world to that, and the people who will thrive in it are those who can change, who can understand the holistic view of all the components that are working together, because the value is implicit in the data, and people say, Oh, what's data science? and I always try and say think of it as data archeology; you just slowly strip away. If you know what you're going to look for, then I'm not interested; you already know it's there, so what are you looking for? How do I find the value that's hidden in that data that's different and that's very different from the common thing? Oh, I want to report on X. I want this if you already know what you want, et cetera, et cetera. what you're looking for is nice It's very much exploratory; it's being able to combine different parts of the information space with what the business is trying to achieve. It's a very different and rapidly changing world. So do I want somebody who was a virtuoso in Python? Why would I even want that? Because it will be out of date, it's one component maybe a while back. No, so the skills that I'm looking for in people are quite different from the skills that the average JD asks for or your HR person expects. The world has moved on; we're not making widgets anymore.

TIM: The skill set you're looking for, you mentioned, then the kind of ability to learn and adaptability—is this more what it's about? Is it just you need smart people who can solve problems that they haven't seen before, learn new stuff because tools themselves are changing so rapidly? Is that it? Is there anything else that you would really look for?

ROBERT: The ability to learn, the ability to question—that is one, because the idea is, and I'll just use the phrase AI. AI isn't about automating the past; it's about reimagining the future, and there are two completely different approaches, and instantly you can pick one and the other. One isn't going to get you very far, so not that it's not useful, but it is the world that we're now sitting at the junction and moving into. It's not automating those processes; it's making those processes redundant. You don't even need them anymore. But lots of companies are just hiring these people, and all they do is automate. What they had was nice, but no, if you can reimagine what you can do with these tools, that's the sort of mindset you look for. and that requires people who are confident in themselves to be willing to learn and have that natural inquisitive sort of nature. Yes, they have to master the tools, but those tools will change every year, so there's no point in going. Oh, I've studied up on this, and that was nice. But it's, and we are talking 12- to 18-month cycles on a lot of these tools, so yes, it is very hard, I admit. So you want people who can adapt, you know, and learn.

TIM: You had a great phrase there where you said maybe you could repeat it because I'm going to butcher it, so you can automate the past so we can reimagine the future. Is that what you said?

ROBERT: Yeah, that's what I said. Yes, that's pretty close. Yeah, and it is about that, and it's very hard to put that into a JD.

TIM: And I feel like then there's a discussion here about hiring because what I suspect is going to happen, at least in the first wave, is companies are going to do the former, which is automate the current way, so they're going to be Oh, we've got all these CVs coming in. I can't possibly read 1000 CVs. I now need to automate CV screening. Oh, we've got too many candidates to interview. I'm going to automate interviewing the data that kind of going to try to just automate the way they're currently doing it rather than maybe thinking, for example, Oh, what are the data points we have never collected about candidates that we don't have any idea about? For example, how are we going to solve some of the systemic issues in hiring? Think about how we are going to hire the best person as quickly as possible. Like, how do we do that? How do you think it's going to go in terms of AI and hiring?

ROBERT: I think you've hit the nail on the head in a sense. Yes, and I'm not saying it's not good to automate some of those things, but it's far better to reimagine what it could be like. I said you could break the person physically down into those components and then look at them in different spaces to find what your candidates are. It's got nothing to do with the CVs, and I guarantee you'll be more successful. It's difficult whenever there is a change; it's like going Oh look, I've got horses, and we need more stables, and I go, I've invented the motor car. Oh, okay, we need more hay, then they just keep going back to the same sort of thing. It's always that way: those who can take the leap and say, Wait a minute, I can see the problem is completely different now. I don't need to automate it; I've just engineered it out. It's not even required. I'm making it black and white, but it is very much that way. Let's rethink you know the CD you know. It's old, and it was only even an approximate method of getting yourself across. I don't know; it's had its lifetime, your curricula vitae. It was good when we all spoke Latin, but now it really is a piece of administration for the sake of administration.

TIM: And if I'm not mistaken, this might be an old wives tale, but I'm sure we've posted about this on LinkedIn in the past. The first recorded CV ever was Da Vinci, and I'm no historian, but I feel like he was alive more than 500 years ago, and it feels like it hasn't really changed much in those 500 years.

ROBERT: Yeah, I think you're right, and it probably goes back before that, no doubt, but yes, you know. We're in a world where you're online; you interact with information differently. Why are we, you know, why are you asking for a piece of paper? Seriously, it makes utterly no sense. But there's an industry that's based on it. That's its only real reason for existing. And it's going to go extinct because they're just not going to change, and people will just start working around it as they already do.

TIM: What about thinking about the typical makeup of talent and HR teams? What impact do you think their skills in data and AI, or perhaps in some cases lack of skills or knowledge in AI and data, or even coming from something close to that background, like it would be pretty rare that someone leading a talent team would even be from, I don't know, a software engineering background or science background or something typically from other backgrounds—is that going to limit how quickly hiring changes? How soon some of these more advanced methods are adopted do you think

ROBERT: Yes, I think it does. When you're skilled in a specific way of doing things, that's the way it's done. It's hard to change. It doesn't mean you can't change; there's the natural resistance to change. Evolution will drive that change. So the best thing is, as I tell people, you either ride the wave or get swamped by the wave, and this wave is coming, so people should be working. Okay, how can I do things differently? And different is not, Oh, I'm going to get AI to read my CV. The CVs—you—and then you look at them in a funny way, and you go Yes, but those CVs were probably written by AI. You have AI checking AI, and I'm not joking; that's the case. Just put your thinking cap on; that cannot be a good thing.

TIM: And that's certainly what we're hearing at the moment consistently: yeah, so these companies are getting inundated with these ChatGPT-written or optimized CVs, which now surely must be even less representative of reality than they were before, and before they were already pretty shit, now surely the sort of truth percentage or truth index of a CV is going to be at an all-time low. In which case, what are we really screening it on? What does it mean? Does it mean anything?

ROBERT: That's the question. I speak and think of each day; it is true, and you look at it and go You have a process for the sake of process, not because it's achieving what it's doing. The world has moved on, and you can stand there with your finger in the dam hoping it's not going to break, but everyone's flowing around it, and those companies that make that move quicker and, you know, take up these skills AI is a minefield of opportunity. Now it's a minefield. So it's not just as simple, but yeah, but no, the fundamental is why are we handling pieces of paper back and forth? That is basically fiction, so you ask me, Oh, can we speed up how quickly we can read the fiction you That's the wrong question. The right question is why are we dealing with fiction, not can I speed it up? Does that make sense, or am I being harsh?

TIM: No, I think you've been harsh but fair, as they say, and it comes back to your original point of wanting to hire people who ask good questions because if you live in the madhouse and you never question why the lunatics are running the asylum, then I guess you'll be in there forever with them.

ROBERT: Yeah, I am trying to make it black and white, and there are shades of gray, but the underlying truth is the CV has had its day in about the end of the 1890s. To be fair, we all know it's fiction; we know it's been written by AI, and we know we're getting AI to read it. It's just ludicrous now in my mind; a good HR person is not a body hunter but is someone who can coax the talent out of their shell to see what it is and to see past it and go, Ah, that's a talent. It is truly when I find—and there are some I've worked with who are talent managers—the word talent, not HR, and they understand, and you can work with them and say, This is what we're looking at. This is what I do; that is the role I see an HR company should be providing, not body shopping.

TIM: And for the good talent managers you've worked with, you mentioned they can coax out the diamond or something like that. In your view, how do they do that successfully? What are they good at?

ROBERT: I think two things: One is they understand what I'm looking for in that person, right? They spend time because it's never a black-and-white thing, and they get a very good understanding of what I'm using the word my for, but for any manager or CTO working, what they think and how they work generally, they take that time and those characteristics. and then when they're dealing with the candidates, they have a very good idea of those initial matches, and it might not be obvious that person has those, so they'll spend time with the candidates chatting through and then realizing, Ah, okay, so really you just build this in the CV so it matches the JD word for word, but you do have these; you realize that there is a lot hidden. So they take the time, I suppose I'm trying to say, and it's at both ends of the scale from what the hiring person requires and what the candidate can deliver, and that's when the AI tools would be very quick and useful, being able to classify a person in a nice way when I say classify, and then you're like, Ah, here are four of those types that generally might fit. So I'll spend an effort on getting to know them, et cetera, et cetera, so you'd use the two together.

TIM: And as we presumably progress to this more AI-driven hiring approach, whatever that ends up looking like, maybe reimagining from the ground up how we do it, where in that equation do you think a cultural fit interview would sit, if at all? Because I feel like my experience of these in the past has been that a lot of the time they sound good superficially, and they sound like they're being done for the right reason. but I'm very dubious that there's really any science behind it at all because they're rarely measured, and it's often about feelings. And I feel like we should try to be doing more fact-based things rather than feeling-based things. You can tell what side of the fence I'm on. I'm wondering, yeah, where do you see these sitting in this new AI world? Are they still in there, or are we throwing them out?

ROBERT: I'm like you; my whole career has been based on evidence-based decision-making, and often that can go against the feelings. I'm not saying feelings aren't good, but like I said, feelings are filtered by your own bias, and you could miss the opportunity or the talent. Having something that is neutral if only it questions what you came up with yourself Oh, my feeling said X, but this is Y. Maybe I should readdress that and look at it; otherwise, you just go on down, so even just as a counterbalance, what it allows is to have a baseline that you can measure against. That's the gold standard in science: having something to measure against in nature, and then you go, Oh, is this what that was? But if you have something that you can say was an X and this was a Y, then eventually you go, Ah, so for that purpose, the lane of always having something that is reproducible. and then measure it against it, and now you can go back and check, and now you can say, Ah, this is what the data is showing, so all the parts are needed.

TIM: for talent teams then in this new era How do they need to redevelop? How do they need to rethink if you're working in talent acquisition now or advising talent? What is coming in the next few years? If you were working in talent yourself, how would you be thinking about upskilling and changing your views on what's on the horizon for them, do you think?

ROBERT: I would definitely want to be familiar with the tools that are out there. Yes, people are using creativity, but knowing exactly what they are, how they work, and what they can and can't do is important because this is a world that you're dealing with, the good and the bad, and you need to understand it. So it's not inflicted upon you. You can say, Okay, I know what's happening here. To be aware of it, that's all you'd have to be an expert in it, but when I mean being aware of it, it's not good enough to watch YouTube videos or read articles on it. No, you need to understand what it does and what it can't do, the good and the bad. And then again, it's getting an understanding because often the person who's making the hire doesn't even know what they want to hire; they don't understand the role that they need. It's not their fault, so for me, a good talent agency or something should be able to go and say, This is the role you think you need; from our experience, these are the roles; this is the future you're asking for; this is the role you think you need; this is what these roles are turning into," and because they're not going to know necessarily, you should be able to help mold the role that they think they need and at least put those options out there. the other way, of course, is if they're hiring senior people and we're looking at executives or chief data scientists or senior, etc., often they don't have anyone in the organization who could even take the interview who would understand it, so part of what I'd expect them to do is say we know a data scientist. We can get one in who can go through the interview process and look at that side of the hire because we know it's technical, so it's like saying, Oh, we want a brain surgeon, but I don't have any brain surgeons to interview them, but I have a good CFO. It's exactly the same; however, a good recruiting agent should be able to go and find we've got two great data scientists, etc. They can be part of your interview process and then pass that information on to us, or you. It's going to be needed because there's such a huge gap; otherwise, it's a blind leading the blind.

TIM: Yeah, and I think you've touched on a really important point that isn't really, as far as I know, solved for in any of the HR tech stack at the moment, which is, oh, what if the person were—what if the role as we've defined it itself is wrong? Like, normally that's almost like a given; that's an input to then the system. Okay, this is what we need. Let's go and find this, but you're right; it's almost one step above that, isn't it? Then the question is, is this the right thing? Unfortunately, I feel like that's going to be quite a stretch for the talent and recruiter teams to then be questioning the hiring manager because, in theory, that should be their domain. They are, in principle, the expert, but if they don't know, then that company's in a bit of trouble, I would have thought.

ROBERT: Yeah, and whenever something new comes along, why would they know, right? It's like we've never had brain surgeons before, but we need some, so a good HR person or talent person would be able to take what their role is the way they've described it, and they expand it and say, Here are the other options. This is what we're seeing in the market: these roles have now been replaced by these sets of skills. It's not telling them what to do; it's saying, If you've only ever seen black and white, I'm showing you color TV now. Maybe this will make a difference. The HR person should have had access to many roles and seen what's happening out there, saying those roles are really what I expect from a good talent person to be able to put out options. Not to be insulting, saying that's not the right route, but saying in these organizations I've now moved to this sort of approach you've got to be diplomatic, not my strong suit, but it's a requirement. We are moving to a different paradigm; we can't use the old stuff.

TIM: Is there an element, then, in looking for people? So we're looking for people who are adaptable, who ask questions, who are willing and able to learn. I feel like there's almost like, how do we measure that? Because I'm not an expert in psychometrics and psychology, but there's the big five personality model where you can survey people and you unpack their personality into conscientiousness, neuroticism, agreeableness, etc. I feel like there is a new factor or a new way we have to assess; I feel like it would be a goldmine to be able to evaluate someone and go, This person has a really high learnability index or something, or is it just come down to how smart they are? I don't know.

ROBERT: I think as we start to collect data on this and create models, we will be able to classify people. Not like the old Briggs and thingy tests and all those, which were all made up, but the data will show us patterns and types. We need to collect it on that, so it's not an easy solution. It's not a what can we do first thing is to be aware of the problem, to be aware of the bias, to know there is something to look for. The next thing is to know that whatever you're doing, collect the data, even if you don't know what you're going to use it for. Start collecting it and building that up. But yeah, it's difficult because the way HR and talent are grown, it hasn't had to rely on the data on science much. The science I've done is pretty basic, but those luxuries are going away. Like I said, a person can be expected to have four careers in their lifetime now with changes in technology and moving so quickly. So how are you going to go? Oh yeah, that's a perfect job for you, or it should be. That might be today, and will you be good for tomorrow as a CTO or as a chief data scientist? My role is to look three to five years into the future, so when I'm looking at people, I'm looking at where they are going to be. Are they going to be around in three to five years? That skill set I don't want to have to keep starting hiring it, so I'm already looking three to five years ahead, so I have to judge everything that we're hiring. I'm people like, I'm not sure what it's going to be like, but I know, all right, that person seems to have a good chance of being able to enter that role in the new world. And it'll be very different from the world I'm hiring them into today. I can promise that within three years time, I would have picked completely different skills to hire that same person. Difficult, isn't it?

TIM: I think it's difficult because, as you say, the rate of changes is so fast, and I think particularly with large language models, it feels like their development has already broken a lot of the hiring processes fairly fundamentally, and companies don't quite know what to do, so there's this screening step where it's okay now candidates are applying en masse with a CV they haven't written; that process is now dead, then there are candidates using it to take coding tests. Hang on, this job is all about coding, but if the AI can do it, then it's almost like the penny's dropping at some point. Do we need the person? Not really. If you're saying their entire job is doing X and now AI can do X, then why do you need the person? And so I guess it's just hard because it's changing so quickly. What about I'd love to get your view on this: Do you view this current rate of change with AI as something we should expect to persist in the next couple of years? Do you feel like the 2026 and 2027 versions of what we're looking at are going to be equivalently

ROBERT: honest answers honest answers Of course the future's coming quicker than people realize. My role is to look into the future and what I can see coming. It's exciting. Most people do not realize what's already arriving on their doorsteps. I'll give an example. You'll start to read and see a little bit about bringing recreating people who have died back and having an AI avatar, et cetera, and it's starting to— it started with pop stars. Oh, I've got ABBA up there as avatars to actually bring back Should we say digital versions of famous people or whatever? Okay, there's just a you can do that pretty straightforward, and all I know, and it's got all sorts of weird connotations, but it is possible in a sense to recreate what a person was like in their past and then to recreate it digitally now at the moment they're doing it for people who have died or relic friends or whatever. It's got a certain creepiness about it, but it is you. It is there. It's not hard to do now if you just move that over. If I can recreate a dead person, I can recreate a live person, so if I take the best brain surgeon or the best school teacher or the best this, I can make a thousand copies of them and have them doing their job.

TIM: And is there going to be a Robert bot out there soon, do you think?

ROBERT: I would not inflict myself upon the world, but it's not a bot. Why wouldn't you reproduce the best schoolteacher? Those schoolteachers who are able to teach me math somehow, and you can capture their characteristics, it's not really—and then reproduce it so that every school can have this great teacher or, but just take care of what they're doing now to recreate dead people can be with little improvement used to catch the essence of the best schoolteacher or the best doctor and then reproduce that so it can be democratized in a sense. That's just a simple sort of potential of what is just around the corner and people are playing with already.

TIM: Yeah, it's amazing, frightening, and exciting at the same time, and it feels like we could have an incredible unlock in so many things, and the mind boggles with the problems we could solve if we could replicate a million Einsteins and have them solving complex physics problems. Let's do it.

ROBERT: Oh yeah, that'd be great, but it's as simple as schools don't have enough good teachers, and kids are trying to learn at home. We can actually reproduce the style and method of Mr. So-and-so or Miss So-and-so, and even at that humble level, it will be revolutionary. Basically, why have a crap teacher if I don't have a good teacher? Most people are learning at home a lot of the time, so you don't really see your teacher hardly anyway, but I am trivializing it, but that ability to take the best and reproduce them in a manner that is not too creepy, and there's all sorts of legal issues and all sorts of things, but the technology is here and getting better. So you could say this is the best HR talent person I've ever had. Okay, now you hire that out.

TIM: Yeah, and I hope we see the replication of the best and not the replication of the worst because that could be the terrifying alternate universe, unfortunately.

ROBERT: Absolutely, it's an interesting sort of world. It's potential is out there, and that's potential we can see. It's the ones that surprise us from left field that I just know will be there, good and bad, etc., but I think it's always been like that. We don't know the power of something till it's too late. But there are wonderful opportunities that it has, so people who are just trying to automate today, I look at them going, You have no idea what's coming tomorrow, do you? And that's what I mean when I say, No, you've got to reimagine the future and grab it before somebody takes it or does something bad with it. In the subject area we're discussing now, it's no one would disagree when I go, Why are you handing me pieces of paper? CDs really send me a telegram; that'd probably be more modern. It is really, when you think about it, utterly ridiculous.

TIM: I agree. I agree completely, Robert. I have one final question for you, and that is if you could ask our next guest one question about anything, what would that question be?

ROBERT: I suppose in the subject area we're discussing, I'd be really interested in what your plans for the future are that we've discussed. Are you just going to let it wash over you, or are you being proactive, saying, Okay, I have a plan, or I'm going to look at it, or I may sit back and wait, or Some people will sit back and see what's there, and some will grab it by the throat? I suppose it'd be interesting to know what they see in the next three to five years because change is so rapid, and we can only guess with a limited view. Like I said, you know, being able to reproduce digital people sounds amazing, but I bet you there are other things even greater that I haven't even An inkling of that will come out of it, some good, some bad. But yes, what's tomorrow's world looking like for you?

TIM: We'll level that question at the next guest, whoever that may be. Robert, thank you so much for joining us today. It was a really interesting and wide-ranging discussion, almost slightly philosophical at times. I learned a lot. I got a new perspective on a few things, so thank you so much for sharing all of your thoughts with our audience.

ROBERT: a pleasure; it's always interesting. Thank you very much.