How Uber is leveraging AI in Recruitment with Elliot Suiter

Thomas Kohler:
Today’s guest, Elliot Suiter, who is leading the EMEA recruiting at Uber.

Elliot Suiter:
How do we make sure that we’re assessing the person and not a GPT answer? Right. We’ve had certain cases already where we don’t know for sure, but we all use GPT enough to know that when we saw certain answers, we were like, that seems like it’s auto generated, it doesn’t seem authentic, it doesn’t seem generic. When you know the person especially, it’s so easy to figure out. So I think if there aren’t tools already, and I’m sure there are, we need to also figure out how to assess people who are using GPT to basically game the system of how you interview. Now at Uber, we do a round of recruiter phone screens, then the hiring manager meets them. Then typically they would have a competency, one or two competency based interviews followed by a case study at the end where you’re expected to pull together ten or twelve slides versus an answer, an exercise. Right. That demonstration of work that you would typically be doing in this role, we find is a really, really good predictor of success and also motivation.

You know, if we send you the case study and you decide, I’m not really that interested in doing it, well, then you’re probably not motivated enough to come and join Uber. Right? But given that you can use GPT to, if you know what the competencies are going to be, there’s no reason why you couldn’t say, think of a really good answer in an interview setting to stakeholder how you would manage stakeholders. Now, that person may never have managed stakeholders very well in their life, but if they’ve got the good answer from GPT, then maybe they look better than they will. So we need to be a little bit smart about how we use competency based assessment from a behavioral as well as a situational standpoint, because it can’t just be what is good stakeholder management, or how would you define good stakeholder management? It needs to be contextualized to let’s imagine that you’re doing this thing at Uber and the restaurant partner says that they want something else. How would you manage that person in that scenario? So I could see a world in which the interview process becomes more like role play, more situational than just behavioral. And what have you done in the past? I think it’s important to get the right blend because otherwise you end up with candidates who just aren’t going to be able to answer the question because it’s too contextualized to Uber or too contextualized to the company that they’re interviewing for. But I think Google and Microsoft now are both in a race to make sure that they have their products like Google Slides or PowerPoint fully enabled with AI. So that you can go and say, create me five slides that talks about X, Y and Z, and then it might be 80% done and then you finish it off kind of thing.

So it’s going to be interesting to see how that comes in more from candidates. I’ve definitely seen it already. I’m interested to see what else is going to come.

Thomas Kohler:
Elliot and I talked about AI intellect acquisition, how Uber is using talent acquisition, and also how he personally trained certain models for talent acquisition purposes or other purposes to leverage AI. And I think it was really insightful because there are a lot of things that I heard the first time when using or leveraging AI and talent acquisition. So that’s definitely a must listen episode with Eliot.

——

Thomas Kohler:
Great to see you, Elliot. We met each other now in person in Amsterdam. Then we had a chance even to work together, and now we are doing a podcast together. And I’m really excited because your passion of AI, intellect acquisition and also everything that you told me upfront was super, super exciting and I’m really looking forward to hear more today. But maybe we start with a short introduction about yourself.

Elliot Suiter:
Yeah, sure, it’s nice to be invited on and it’s been great to get to know you over the last year working together, predominantly in Germany, and very excited to talk a little bit about AI and the intersection of that and recruiting and candidate experience. So as a very brief introduction, I’ve been at Uber now for about four and a half years. I currently look after the talent acquisition function for EMEA and the delivery specifically. So we do about 350 hires every year across sales, operations, strategy and planning for various different lines of business. Obviously, I’m guessing everyone will know Uber eats, but there are lots of ancillary revenue streams like ads, Uber, grocery, and a few others as well. Before I was in this role, I was looking after the executive recruiting function. So slightly different level of role that we were focused on. And I was a recruiter and a sourcer when I was in executive recruiting as well.

So it gave me, it was my first chance to kind of recruit inside a company internally. Before that I was at Egon Zender, which is a global executive search firm, which is where I guess my formative years came in. So I was focused on hiring chief digital officers, chief technology officers, but also CEO’s of smaller digitally focused businesses. And I guess that’s what led me to Uber, is that I’d recruited in exec for three years at Egon Zender and worked with some brilliant companies. But I wanted to do that for one company where I was going really, really deep into the culture of that business. And I felt like I was able to have more impact longer term on one company versus getting all the different learnings in, I guess, what is kind of a consulting environment for talent. Before Egon Zender, I spent two years at Boyden, which is also a global executive search firm, doing stuff that’s very similar, CIO work predominantly and then something completely different before that, which was running my own business for a year. And I often look back on that and see that as really formative as well because it kind of taught me how lonely it can be.

So I know that you’re a founder of your own business and it can. Whilst it may seem glamorous at times to people, externally, it’s extremely lonely at times. And you have a. The highs are really high, but the lows are also really low. We run an online website for tailor made suits. I learned as well through the process what I’m good at, what I’m not very good at. I think I’m good at sales, but I think I’m better at selling services like recruiting or opportunities like recruiting than I am at selling specific product. But who knows, maybe eight, nine years on now, maybe I’m a lot better at selling products as well.

Thomas Kohler:
But also the passion on AI also. Why not an AI product at some point?

Elliot Suiter:
Possibly, yeah. I think just going back to that early time, the technology that was being used for the website was, I want to say, fairly advanced for the time. It was, you would go onto the website, you would measure yourself at home, you would pick from an array of different fabrics and designs, and it would build the suit for you in real time. I think there’s probably 101 websites that do something similar now, but at the time it was super interesting. And what also caught my attention, or what I really enjoyed was the social media aspect of how you build your business. So using different channels, either for paid advertising or, or just for social media, tracking organic content. Again, this was at a time when, I guess early, it was about 2012, 2013, and that was when it was really starting to proliferate on Facebook, on Twitter, on Instagram, and all those kind of different channels. I found that really interesting.

Thomas Kohler:
Cool. And what is your touch point now with AI in talent acquisition?

Elliot Suiter:
Yeah. So I think for me, the arrival of Chachi BT especially has obviously transformed how we’re going to think about work overall. But I think there are definitely industries in which it will be more disruptive, and I think recruiting is certainly one of those. I guess our disadvantage here as a human being is that our brain can’t store a lot of information comparative to AI for a long period of time. And we can’t recall it as quickly either. And I guess because of this, it limits how efficient we can be. And it also limits a little bit, I think our decision making process, it can become a bit flawed, it can become a bit biased if you look at things like recency bias and that kind of thing. So for me, the very first time I used Chachi Bt especially, and there’s obviously lots of different AI tools out there, but the first time I used GPT, it was a real light bulb moment.

I could see instantly how it would have an impact on the way that we work. And obviously we all know it hasn’t even scratched the surface yet. And there’s lots of things for us to consider as we implement different ways of working using AI, but some of the ways in which we’re thinking about using it at Uber and we have a chat GBT for enterprise license. Interestingly, we also sublease or, sorry, I think lease out our offices in San Francisco to open AI as well. So I think we have a fairly good relationship with them. I think it’s safe to say. I think that creating new search strategies and understanding different market trends, developing questions for interviewers to ask when they’re assessing for certain competencies, and we’re also looking at different tools that we can add on top, like Zoom AI assistant. So having tools that take notes, summarize those notes with transcripts and then comparing it to the role spec as well, those are some of the early things that we’ve started thinking about it for.

Obviously, you can also build your own GPTs now as well. So as an experiment, I built a GPT that would allow us to kind of summarize all of the interview notes on a specific candidate and compare it to the role spec so that we were able to, I was able to running the debrief to be able to say, hey, these are the things we’ve agreed are the must haves and the nice haves. Based on all of the information that you’ve provided from the interviews, it seems as though the sentiment is that we are doing well or matching well to these must haves. But the two things that we’re not really focused on or haven’t come off as positive are these two areas. So I want to focus the discussion today around these two areas. And I think most critically, that’s where AI is still a little way off. I think being able to fully replace a recruiter, it’s that human discussion piece, it’s the creativity piece. I’m sure there’d be lots of people who disagree with me and say, yeah, but that’s coming really, really soon. And maybe it is, but at the moment, I can’t foresee that discussion being replaced with AI.

Thomas Kohler:
Definitely, I agree. So I think that’s something. What I also see is that there are a lot of things like writing a scorecard, which maybe AI can really do much better, and then you can focus on the interviewing piece, and then you are way more in attention and can provide a better experience. And I don’t know if for a human, it’s a good experience to be interviewed by an AI versus a real person. Right? Yeah.

Elliot Suiter:
And that’s the whole point, right? Is that my view on this is that AI will replace more of the menial tasks, the things that, quite frankly, recruiters or sources in general find less stimulating in their role and that time saved. You can focus on the candidate experience. Now, we haven’t explored the use of AI in the form of, say, chatbots to improve the scheduling process or the candidate experience yet, but I think there also are cases where, for example, let’s say a candidate wants to know more about the culture of Uber after their first call with the recruiter. Now, we send all candidates a preparation pack for their upcoming interview, but let’s say they missed it or didn’t see it or whatever. Let’s say that they also sent this message at 10:00 at night. Well, they could get an instant response with a pack that’s focused solely on candidate experience. Now they’ve got it when they want it, and they’ve got it instantly. So I do think there’s a case where, again, for those more, slightly more menial tasks, it’s going to radically improve the candidate experience.

Thomas Kohler:
In case you like my show, please subscribe. I would really appreciate it. And I think that’s also interesting, what you just mentioned, because I did last year an investment in one company in the HR tech field, they are focusing on blue collar workforce and building an HR infrastructure for them, or let’s say a holistic tool because it’s mainly the big number of workforce if you look at it, versus all the white collar jobs. But all the tools are built for the white collar jobs and not for the blue collar jobs. And what they started with is like having an intelligent, let’s say, technology or AI powered way to just keep the data up to date for payroll. That because payroll is a big bottleneck and a big pain. And the biggest pain because payroll is getting a pain is the data is not accurate or up to date. And the easiest way to – or – and then they focused on starting with the easiest way for a blue collar workforce worker to update and double check. Is the insurance number correct? Is the address still correct? And they have, of course, a lot of data for, from, let’s say, trial piloting customers. And they just analyzed. Okay, what’s the average time when some of these data points will change? And let’s double check and just send a WhatsApp to them and ask in a chat like they are used to it as a user, is it still up to date? And if they get a yes, okay, data is up to date. If no, then they send it and they update the system. It’s not going through HR manually, right?

Elliot Suiter:
Yeah.

Thomas Kohler:
So that’s also a cool use case, I thought.

Elliot Suiter:
Yeah, definitely. But what you touched on there is, I think the more exciting part of AI transforming our industry, which is that it’s not going to be one tool. A little bit like, I would say LinkedIn really changed the game for recruiting, say, 20 years ago. That was one tool that really shifted things. Right. The way of working now is completely different and we’re very reliant on LinkedIn, also other tools and channels. But that’s, you know, that’s a, that’s a primary tool. I’m not sure that chat, GPT or other tools like it will be a one stop shop or a finished solution for how recruiting improves.

I think it’s going to be, as you said, almost every company adding a layer of artificial intelligence to existing or to new products. So, for example, with Zoom, right, I think they’ve done an incredible job of making merger and acquisition over the years and then integrating those businesses into the Zoom product so that now they are at the forefront of, well, everyone uses Zoom, so how can we optimize it using artificial intelligence? We can use summary transcripts. We can allow people to catch up on the meeting. If they join ten minutes late by giving them a summary of what’s happened, we can. I mean, I’m not selling Zoom here, but, you know, I think lots of different companies will add that layer of artificial intelligence. Uber is also a great example. I think we’ve been using AI for probably eight to ten years in its earliest form to do route matching. It’s not something that we, I think we’re singing and shouting about in the earlier days, but as the AI industry or the AI wave has come over the last twelve to 24 months, it’s definitely been something where we as a company are a little bit more like, well, we’ve been using AI for a long, long time and we’re actually quite good at it.

There’s more use cases for sure. We just launched a chatbot within the app so that you can engage with the bot to suggest what you might want for dinner, for example. No, I don’t really want italian. I liked this italian last time. I’m feeling something more like this. What would you recommend? And that dialogue back and forth should give you certain suggestions. So that shows you how early on we perhaps are. But as it relates to recruitment, I think that there’s going to be a lot of different tools out there that just add in a layer of AI.

Thomas Kohler:
And also, I think what you have as an asset, as an organization like Uber, is you have a lot of data, not just maybe on the product side, but also on the recruiting side, because it’s just so much volume. A large organization, I think this can be really utilized to build your own layer of AI tools based on just the data you have. And then it’s something very Uber specific. Let’s say what maybe other organizations with not that amount of data can do. So do you have any potential in that?

Elliot Suiter:
Yeah, I think one thing we haven’t yet tapped correctly is predictive analytics. So, as you say, like, what is the length of time that people are typically in this role? And how can we maybe better predict when someone’s going to be ready for a move and try and manufacture a move internally, rather than someone starting to look externally? For example, we don’t use predictive analytics to say, we hired x number of people into this type of role at this level, and typically they’ve gone on to do X, Y and Z thing that would be really powerful for us to have in recruiting and attracting talent, to be able to say, hey, just so you know, for this role that we’re recruiting for, over the last two years, we’ve hired this many people into it, and x percentage have gone on to be promoted, or x percentage have gone on to do this type of role. If it’s not something where you feel wedded to it or you want to grow in a different direction. Actually, there’s this many people have gone on to do something in a different part of our business. I think one of the things that people always crave when they move company is I want to feel like I’m going to grow at that company, whether that’s personal development or whether it’s career trajectory. They want to be able to see and know what the steps might look like two, three, four, maybe five years into the future. I think beyond that, it’s always going to be a little bit difficult. But I know that when I joined Uber, it was with a view to, okay, I’m going to come in, I’m going to do sourcing, which is what I know and what I think I’m good at.

But ultimately, I want to be a recruiter. Ultimately, I want to lead teams. Beyond that, let’s see what it is. But if my recruiter at the time had been able to say, just so you know, the sourcing function at Uber, we’ve hired this many people and this is what they’ve done afterwards, that would have given me a lot more confidence that it was the right place. Now, I guess you could argue the other way as well, because if you haven’t got a good track record, maybe you don’t want to share it as openly. But I think just over 30% of the hires we made last year were internal moves. And that’s something we do pretty well at Uber. So I think it’ll be something we’ll want to shout about.

Thomas Kohler:
Definitely. It’s great. And I also met some Uber employees, managers, whatever, along my career. And I think everybody really always appreciated the experience, the track record, the relationships they made there, and also the insights they got or the. The knowledge they got right. And I think that’s something what, of course, can be a big, big asset. And also for career planning, this is also something I think with now, companies recently started doing more that they look also at the internal workforce more holistically. And I think this is where AI can be a real layer of adding a certain structure to thinking because I think some people in certain roles, they maybe are not trained in strategic thinking. They are maybe not trained in analyzing situations in a very structured, objective way, but have maybe a very good intuition and are just right with their gut feeling because they did things so long in the past, and that’s then maybe intuition that is powerful and adding the layer of a certain strategy or structure or analytical approach to thinking through AI and processing certain, I don’t know, prompts. This is, I think, a very powerful combination because then everybody has the possibility to operate at a certain level.

Elliot Suiter:
Yeah. Do you do foresee, obviously, you and I work very closely together. Do you foresee also any sort of skills gap or skills shortage with sources and recruiters as tools like GPT, but also AI enabled tools come to the market?

Thomas Kohler:
I think for that I’m a bit too far away. But what I see as a big gap in skill sets, what differentiates maybe an outstanding recruiter from an average recruiter? And then especially with my model, right. If you hire us, and I would provide just average recruiting support, you would ask yourself why to bring on an external supporter or consultancy. Right there. I see that if people are not trained in certain ways of thinking and analyzing situations, they fail in certain situations, like with bigger organizations, demanding stakeholders, and maybe not being able to anticipate or resolve certain bottlenecks, like you don’t get the attention you need or you don’t get the responses you need, or people are moving too slow and the urgency is there and so on. If you’re not able to think outside of this box, like your view, and make sure, okay, to think in the perspective of your audience and then solve problems from different angles, that’s a skill set that is very demanded, but also, nobody has it just by being born to it, right? This needs to be trained. And I think that technology can really be an asset there. But the question is then how do you get the data to proceed or to solve certain situations that are maybe individual, unique and so on, right.

Because at another client or with another stakeholder, it can be very different viewpoints. And the question is, how do you gather the data that you can process them? So I think that’s something what I would see a big asset in the future to fix a skill gap in recruitment maybe, or in sourcing, but my imagination would not tell me how this would be possible.

Elliot Suiter:
Do you think you could use AI then to do training and learning for either new joiners or training and learning to upskill like a recruiting or sourcing workforce?

Thomas Kohler:
Oh, definitely. I think that’s possible that you say, hey, these are the best practices to resolve certain situations and they are recorded, and then I don’t know that the technology is trained on it. And then for new starters in onboarding, it’s like, okay, like Metaview or something is sitting in the Zoom call and then recording the interviews or listening into the interviews of stakeholder calls, of recruiter interviews, whatever it is. Right. And then just analyzing as you did it with Uber. That’s actually what we actually want as a best practice or an ideal scenario. This is how you did it. Here would be the gap in behavior, and this is how you could train it. Right. So that’s something that could be super exciting.

Elliot Suiter:
But one of the words that you used a couple of times there is like training something like training a model, training a person. I recently pre ordered a rabbit r1. Have you heard of this or not?

Thomas Kohler:
Yes, you told me last time and I looked it up. Yes, I will also order.

Elliot Suiter:
What did you ordered one?

Thomas Kohler:
Yeah. So, I mean, I did not order one because I needed to do, let’s say you from it was, I think, us shipping. And then I shut down my laptop and it’s on my to do list. Then I’m back in Berlin and I have a bit time for admin. I will do it.

Elliot Suiter:
Okay. You should. I think it’s gonna be a one of the. I ordered two because I want to play with one and see what it’s like. And the other one, I’m making a bit of an investment bet that it’s going to be one of those first iPhone or first MacBook moments and that I’ll be able to sell it for a lot of money in a few years. But I guess for anyone who doesn’t know what the Rabbit R one is, go obviously have a look at it on YouTube or wherever, it’s difficult. I’ve spoken to a couple of people about it now, and it’s very difficult to describe in a sentence or two what it is. But I think the mere fact that we now have a hardware device that is leveraging different platforms like ChatGBT and anthropic and others, I think the fact we have a hardware AI device that isn’t made to replace your phone, they were very clear about that in their keynote, suggests that we are, again, very, very early stages of what’s to come.

Someone said to me as well once, so you know everything that’s there now, chat GBT, and you’re really excited by the technology. Imagine what they know who are building this, of what it’s capable of, that you have no idea what’s coming yet. And that really made me kind of sit up and pay attention because you think, God, this is really the beginning stages. But the rabbit l one again, just as, I guess as a very brief summary is an AI hardware piece that you can. It’s a bit like, I would say it’s kind of like Siri on steroids. We all know, like, you ask Siri to book you an Uber and it will just show you where the app is, and you click on the app and then you go and do it. The rabbit one real’s kind of USP is that it’s built on a large action model, and if you ask it to do things, it would connect to the apps that you have, thus not replacing your phone, but it would take actions on them as well. One of the other selling points that they had in the keynote was that it, you can train it.

So just coming back to the point that you made, I think also with the GPTs that you can now build in chat GPT, for example, I think we’re going to see a lot of people building their own version of something and then training it and making it better and better and better and better versus just leveraging what’s there. So, for example, not that I’m fully aware on the data privacy implications of this, and we certainly haven’t done it yet, but if, for example, we uploaded a huge library of job descriptions as it relates to different roles at different levels and use that as our repository and then just typed in, can you create me a role spec for this level and this type of role? Maybe that’s one outcome. If you take it a step further, you might then say, here’s a cv of a candidate that just applied. Which role would they fit best at Uber, for example? And then you would have to train the model over time to say, I put this candidate forward. And actually, yes, they were a good candidate for that role. So you are correct. You’ve probably seen as well chat GPT often says, was this a good answer or not? Because they want to continually train it. Right with the rabbit r1 piece, I think again, in their demo they showed the camera videoing someone doing something on their laptop to teach it and train it.

And then later they just asked them to do that thing and it did it because it’s been trained on how to do it. So I could foresee a world for sure where you are filming, perhaps how you search for talent on LinkedIn for certain roles, and that you then step away and later go back and say, hey, can you find me some candidates for this thing? And because it knows how to use LinkedIn, because you’ve trained it, it knows where the buttons are, it knows the filters, it knows what works and what doesn’t. You no longer have to figure out how to source. All you have to do is say, this one’s good. This one’s not good because. This one’s good because this one’s not good because I think we’re going to get into a rhythm maybe in five or ten years where our focus, our skill, if you like, is being able to say what is good and what isn’t. Which brings up the question of how satisfied are we going to be in our jobs if we’re just going, yes, no, no. Yes.

They’ll obviously be things that I say. Obviously I need to be careful with that word. There will probably be things that AI can’t replace within the recruiting process, and hopefully the things that it can replace are the things that we are less stimulated by anyway. But as someone who did sourcing for six years, I love the hunt. So I hope AI doesn’t replace sourcing too soon.

Thomas Kohler:
Yeah, definitely. And I think also there is a lot of sourcing for maybe very unique situations which you just do once in a lifetime, right? Yeah. If I just look back. Okay. There are a lot of volume or repetitive tasks and searches, of course, going on, but there are also a lot of situations where this was a very special, unique situation because a company with that business model, in that market at that stage needed exactly to solve that problem with research. Right?

Elliot Suiter:
Yeah, yeah, yeah, yeah, yeah.

Thomas Kohler:
And what you said made me remember, I think keynote also from the founder of OpenAI. What I heard that soon there will be several 1 billion so unicorn companies that are single men, women, individual businesses.

Elliot Suiter:
Yep. Well, you’ve got more than one person, so it’s not going to be you. I guess you’ll have to.

Thomas Kohler:
No, it’s not. It’s not coming me. And it’s also totally fine. I’m very happy with working in the team, but maybe for the foreseeable future that this might happen sooner than we all expect. Right? Yeah, maybe. The next founders of unicorn companies, they don’t need to raise money and find the right co founder team. They maybe just be smart with AI and leveraging technology and having some kind of knowledge and advantage that they see something as you do. Maybe, Elliot, it’s you, but others don’t see it and just test it and get it out there virally and whatever we know.

Elliot Suiter:
Yeah, it reminds me as well of, I keep going back to the rabbit r1 piece, but there was another interview that he had done, and he said, like, we are a very, very small company, and so we don’t really have a big customer service organization. And I was seeing all of these messages coming in regarding, like, customer service. And I just thought, well, if I can train the Rabbit R one to synthesize the main problems that are coming in, like, I didn’t have time. This is him, obviously, as the founder, I didn’t have time to go through all of the comments and respond to them all. So I used the rabbit to understand what all of the problems were, summarize those problems, what were solutions to those problems, and if I think those are the right sort of solutions, we can now build it so that we automatically respond with the right thing. It’s just fascinating. I think, for me, one of the skills gaps that we have, and I hold my own hands up for this as well, is that we’re going to need to become really good at asking the right questions. It almost reminds me a little bit of the, you know, the scene in irobot where Will Smith is constantly asking questions of the.

I think it’s the artificial intelligence hologram. And when he asks the right question, the hologram goes, that is the right question, and then you know where to go next. So maybe it’s just us being able to ask better questions.

Thomas Kohler:
Yeah, definitely. And I think I wanted to say something finally, but I forgot. Yeah, doesn’t matter now, you.

Elliot Suiter:
I was going to ask you a quick question as well, which is.

Thomas Kohler:
Now I got it.

Elliot Suiter:
Okay. Sorry, the rabbit. Yeah. 

Thomas Kohler:
Yeah. I think that this is something what a lot of people just don’t leverage too much to do exactly what you did with the rabbit to leverage it. For instance, there is information, like a lot of information, that is coming in unstructured. And often the responsibility of, let’s say, a manager or a consultant is really simple. Just gather all the information from all different sources of information and structure it in a way to summarize it for a person that’s perceiving the structured information to be able to take some actions, because unorganized information is being organized and has maybe an recommendation on how to take action on something. And this takes a lot of time.

And I think people who are trained in doing so, like a McKinsey consultant, they’re getting fast with it. They just have this thinking model in their mind. But there are a lot of people out there that are never trained on that and never emphasize it. But with something like this, they could really get an assistant of structuring information, gathering information, or maybe gathering needs to be on them. And then it’s up to you, as you said, to decide on, okay, based on that, what would be the best recommendation? Based on my intuition or my experience, this is what I just wanted to also highlight again, because I think what you just mentioned there, it’s really powerful and I was not aware of that and I will definitely check that out.

Elliot Suiter:
I think one of the other areas that we need to be careful on is, okay, so if you’re going to run an interview process and you have, and everyone has access to these tools like GPT, how do we make sure that we’re assessing the person and not a GPT answer? Right. We’ve had certain cases already where we don’t know for sure, but we all use GPT enough to know that when we saw certain answers, we were like, that seems like it’s auto generated. It doesn’t seem authentic. It doesn’t seem. When you know the person especially, it’s so easy to figure out. So I think if there aren’t tools already, and I’m sure there are, we need to also figure out how to assess people who are using GPT to basically game the system of how you interview. Now, at Uber, we do a round of recruiter phone screens. Then the hiring manager meets them.

Then typically they would have a competency, one or two competency based interviews, followed by a case study at the end where you’re expected to pull together ten or twelve slides versus an answer, an exercise. Right. That demonstration of work that you would typically be doing in this role, we find is a really, really good predictor of success and also motivation. You know, if we send you the case study and you decide I’m not really that interested in doing it, well, then you’re probably not motivated enough to come and join Uber, right? But given that you can use GPT to, if you know what the competencies are going to be, there’s no reason why you can say, think of a really good answer in an interview setting to stakeholder how you would manage stakeholders. Now, that person may never have managed stakeholders very well in their life, but if they’ve got the good answer from GPT, then maybe they look better than they will. So we need to be a little bit smart about how we use competency based assessment from a behavioral as well as a situational standpoint, because it can’t just be what is good stakeholder management or how would you define good stakeholder management? It needs to be contextualized to let’s imagine that you’re doing this thing at Uber and the restaurant partner says that they want something else. How would you manage that person in that scenario? So I could see a world in which the interview process becomes more like roleplay, more situational than just behavioral. And what have you done in the past? I think it’s important to get the right blend because otherwise you end up almost.

You end up with candidates who just aren’t going to be able to answer the question because it’s too contextualized to Uber or too contextualized to the company that they’re interviewing for. But I think Google and Microsoft now are both in a race to make sure that they have their products like Google Slides or PowerPoint fully enabled with AI. So that you can go and say, create me five slides that talks about X, Y and Z, and then it might be 80% done and then you finish it off kind of thing. So it’s going to be interesting to see how that comes in more from candidates. I’ve definitely seen it already. I’m interested to see what else is going to come.

Thomas Kohler:
I think that’s perfect. Final words and a nice summary. So thanks, Elliot, so much for your time, and I really enjoyed it.

Elliot Suiter:
No worries. Thank you.