Episode 37

Max Ehrlich on Harnessing AI and Diversity for Efficient and Inclusive Hiring

TIM: Max Thank you so much for joining us on the Objective Hiring Podcast. Great to have you here today.

MAX: Thanks a lot for the invitation. I appreciate it.

TIM: Well, I would love to start with everyone's favorite buzzword topic at the moment, AI. If software was eating the world, I don't know what AI is doing to it, but something pretty amazing, and I'd love to drill down on AI in hiring and get a sense from you if you've already experimented with any kinds of AI in the hiring process, maybe dabbled anywhere. In particular, I'm interested in the screening stages of the process, but if not, certainly anywhere in the process would be interesting to discuss.

MAX: I actually know we are experimenting a lot with AI, but not in the field of hiring itself at the moment. We are trying to keep our hiring process as short as possible, so we only have three process steps until we would give out a contract or not, and therefore I can imagine AI to be a huge, huge efficiency booster to really speed up the process, right? especially as you said at the beginning of the process itself, that's a place where you have most of the candidates, where you have all information in a written format in the CVS, where you don't have any human interactions before you make the first decisions if you want to invite somebody or not. So I really, really, really am very optimistic that this is a field that won't wait too long until, in the industry, there are good common practices that we can use to make the first decisions of yes or no, and with that, hopefully also being faster than our competitors and other companies. and so making the deal faster than that

TIM: Yeah, I think the opportunity is huge, and you touched on one of the dimensions, their speed, which is definitely underappreciated. Like any process that's manual, at some point the human is going to be bottlenecked to that. Even an example as simple as a candidate applying at 5 PM on Friday afternoon, at best, someone in talent or whoever's doing the screen is going to read that at 9 AM on Monday if you're lucky, but it could certainly be much longer than that. I even think about other dimensions as well, like at the moment, explainability is pretty minimal at the screening stage. Like you would in most ATS and HR systems, you don't really know who had rejected them or why or the exact rationale that they took because it's kind of like a slightly subjective yes-no decision. So I feel like there's a lot of upside even in that element for AI to be able to explain why it did what it did.

MAX: I agree a lot, so especially for the candidates themselves, as you said, it's that they often just get a yes or no standard message or something like that, but let's just imagine that you have the time, or I can provide all this information and can tell every candidate why something seems to not be a match and what they could do differently or which kind of tips you would have for them. This would change the market immensely, I think.

TIM: Yeah, I think so, and at the moment I feel like this is done in the way screening is done. I think I feel it is so bad that the upside potential for AI to improve it is astronomical, and the general fears around AI and its negative consequences I feel like are almost reasons for getting rid of the current process. as people would say Oh, AI could be biased, which it could be. Hey, the current scenario is biased. The current human-based scenario of screening the CVs is very biased. I'll give you an interesting example: so there have been all these kinds of studies around the world that apply to various roles with CVs, groups of CVs, the only difference of which are the names on the CVs. and so they're trying to test if there's discrimination against a certain subpopulation. There was one in Australia a few years ago that Sydney University did where they had To cut a long story short, three sets of CVs: the first set with first and last names that were Anglo names, the second set that was Anglo first names Chinese last name and then the third set, which was Chinese first and last name and they just applied to thousands of different roles at different companies in different cities and measured the rate at which their CVs got callbacks. The first group got a 12 percent callback; the third group got a 4 percent callback. So, all else equal, if you apply for a job in Australia with a Chinese name, you've got one third the chance of getting a callback. So that's the current scenario. Surely we can't do worse than that. What do you reckon?

MAX: Yeah, I totally agree with that, and yes, AI will be biased because it will learn from the past, right? It will learn from previous CVs and previous decisions that were taken manually, and everybody, every human, has some prejudices, and they don't know anything about what they believe somebody else is coming from or can do or cannot do. and this I think we also have a high chance in doing a change with AI by really giving external factors and saying Hey, please don't look at things like the name or the picture. Don't look at it, okay? What was the share between males and females in certain positions or whatever? But really, take this as just a subjective point and nothing that you need to or can make a decision based on. So I really think that this can help us if we use it the right way to really be fair and equal to everybody.

TIM: I wonder if then the cool unlock, then, to eliminate or at least reduce the chance of that bias, would be almost like a two-step process so a large language model could easily—you could ask it to, okay, extract for me the skills and responsibilities and experiences from this CV, step one, then match these against the job description. So by the time it's doing any kind of scoring, it's not thinking about the type of candidate it was. It's not trained on like the famous Google example from 10 years ago was. You know, we built a model to look for the best candidates and just kept selecting the ones that had you kind of eliminate that problem by just focusing on the already preprocessed data. Maybe then that might get rid of the potential for bias.

MAX: Yeah, I like the idea, and I think this is very well possible as well, so I'm looking forward to seeing solutions like that.

TIM: What about on the candidate side of things? Like, have you already seen candidates using some of these tools? chat GPT, either for the CV or during the testing stage or even the interview stage

MAX: I have not seen it or nobody has told me that they are using it but I'm very sure they they do if not I would be very surprised I would do it myself I mean Chat GPT or AI is just all around us right so also in our daily business we should all try to use it as good as possible so that we can be more efficient that we can get more information in a shorter period of time and that we can focus on the more individual parts that that are very specific to the company's problem or to the application itself So I would really love to see candidates using ChatGPT as they also do in their daily business to just improve the way of how they're writing the CV, how they're answering things, and so on, so I think we just have to integrate it in our way of thinking and not only in our daily business and work but especially in the hiring process. and why not use it when results will be better or when you can write 15 applications where in the past you could only write five in the same time, right?

TIM: Yeah, it would seem perverse to ban the use of it when you would expect them and need them to use it in the job. Indeed, if they didn't, they'll be doing themselves a disservice and be inefficient compared to someone who is using it. What about is there any limit to that? Like, is there any bit of the hiring process where you would rather candidates not use ChatGPT? For example, in a live interview, would you find it uncomfortable if they were sitting there switching between an LLM and talking to you?

MAX: Yeah, so I think there is for sure a bar where it doesn't make sense to use it anymore, right? So I would also expect you not to listen to ChatGPT right now and just say what they suggest you to say, but really have a conversation between both of us, and I would expect the same in an interview process that we just talk openly to each other and say, Hey, this is what we think or what we don't think. and saying, Hey, I don't know it; I would Google it or I would ask ChatGPT." But yeah, I think this is for sure a very fair point at why not to do it. Maybe, who knows, maybe in the future the standard will be that we communicate by getting real-life information in our eyes or head or whatever, but then it's fair to do, but I think we should just not behave differently in an interview process than we do in our daily life during work.

TIM: Yeah, and I think for candidates, I would certainly not recommend using ChatGPT in a live interview; that has got to make it harder, not easier, if you're trying to read the answer of a large language model and concentrate on what someone's saying and listening to them and paying attention and answering. Oh my God, as if an interview isn't stressful enough, that just sounds like a nightmare to me personally. What about thinking a little bit further afield, like over the next year or two? Is there any bit of the hiring process that you think AI will be used in that there would be like a clear and obvious improvement or use case? Like we've touched a little bit on the screening step—is that where most of the value will be? or could you imagine it later on?

MAX: I was thinking about this question after our last short call, and I really think that this will be the first part where I can play a big role because of the reasons I said before: it's a lot of data coming in, a lot of readable data. The first decision has to be taken without human interaction anyway. but looking forward I think there's such a huge potential in AI for the hiring process, especially when I'm thinking about how today our traditional hiring process works. It's somehow insane. You're writing a job ad; you just post it out to everybody in the world, and you hope that the right candidate at the right time looks at the job ad and thinks Hey, perfect timing! I take it how likely is it that it's not the—this is why today we're going out; we're looking on LinkedIn or other pages to see, okay, do we find a good candidate? Can we proactively ask them if they maybe want to join us? But then you're thinking this forward towards AI: why not have a system where you say as a company, okay, we have this job ad, we have these requirements, we have these wishes, and give me a preselected list of 10-20 candidates who are a good fit? to the chop itself but also likely to switch because of other parameters that you can maybe check how long somebody is in a company, how the company is doing, other layoffs recently, and so on, and when we can shorten this up, this can be a super, super big step not only for the company because it's easier but also for the candidates because they don't have to take 20 calls with headhunters or any messages or whatever; they just get preselected job ads, ideally at the right time. So I think that's a lot of improvement in terms of efficiency out there.

TIM: Yeah, for sure it's interesting to think about where this is going to go because I feel like a bit of the problem at the moment is maybe a lack of high-quality available data to do that kind of matching. Like you've got a candidate's CV and a LinkedIn profile. You make a good point there about, like, other signals that could be collected about the company layoffs. You're right. Maybe also, yeah, you could predict the likelihood of being interested in a move based on how long you've been in a job that's interesting so that's available, but then there's all the things that a candidate would get asked in the first interview or the second interview. That data is all closed off and hidden. If there was some tooling that set as an intermediate level where it was you were maybe as a candidate proactively answering here's the top 20 most common questions, it is going to cover like 90 percent of all interviews. If that data was somehow sitting out there somewhere, maybe then there's an opportunity for matching. I don't know; I feel like there's just something missing in the data quality at the moment in terms of what's available. What do you think?

MAX: Yeah, and I totally agree, and maybe the answer is a platform that is needed in the middle. I mean, it's similar to when I'm going to the doctor; whenever I'm at a doctor I never was before, I have to give all my information again and all the history and so on. This is starting to become a platform so that you don't have to do the same again, and the same happens in some hiring processes. I mean, you come to another company, and they ask exactly the same questions. What's your past? What's your motivation? What was the business for you? You did well. What were good things, bad things? Why do you want to change, and so on? Just think about a platform in the middle where you just have to answer all those questions once, and they can be used and can be shared with all companies that you want to talk with or that you want to share it with. This could save so much time and increase the quality because you can use this time to answer deeper questions and to really, really be sure about whether this is a perfect fit and not just talking three times again about the same CV that you have and so on, and from that standpoint, I think It's looking from miles and miles away; this is just how it currently works.

TIM: Yeah, it's amazing to think of the inefficiency of God—the amount of repetitive questions and answers that have been given in interviews is staggering, so there's gotta be some big unlock there, I would have thought, in the next few years to make this more efficient. What about the bigger picture? Like what for you have been the main pain points in hiring data people in the last year or so? One of the ones I hear a lot about at the moment is like a high volume of candidates, maybe candidates using chat CPT on CVS, those kinds of things. I'm hearing a lot; is that something you've experienced as well?

MAX: Yes, for sure, it's the one thing that the amount of candidates and the other one is also the quality of candidates, and I have the feeling with data being somehow still at the beginning, it's not an industry that has been there for 50 years, and it's still not there that you know, okay. In five years we will use the same tools in the same way. This is not how it's working, right? So we are at a phase where we have far more challenges than we have potential employees that could work on them, and so it's very, very important that we as a company can convince the best talents. to join and to jump on the spot, and I think with that a very crucial thing is that you really can find out who is the right candidate and how can I make sure that we don't see after the probation period or whatever that this is not the perfect fit, and therefore we really have to think about how we can openly Talk about what we as a company can deliver and what the role can give the candidates, but also the other way around: what the real vision is that the candidates have. So that we make sure that we don't tell each other what we think the other party wants to hear but rather openly and honestly talk about, okay, we are growing; we want to invest as a company, but we are not there yet; these are the things that we still have to do and to learn, and we have to go with us through the stage, which will be maybe hard and challenging at the beginning, and it's not always like Hey, the documentation is there everywhere, and all the tools are there already, and the processes work now, and if a candidate hears this and then believes this, he or she will be disappointed because truth will look differently. right, and also the other way around: candidates often try to answer questions like, Hey, this job is perfect because what you write here and here and here is what I want to have. Yes, maybe, but maybe it's not, and then just tell it and say, Hey, these are the things that I hope are hidden there and that I hope that I find there. Then you can still discuss and see, Okay, so something that we will find common ground in the future, and if not perfect, then concentrate on other jobs and other areas where you can move, and you don't lose six months until you find out, Okay, it's not working as I would hope.

TIM: That's a great suggestion, and I couldn't agree more that hiring could be drastically improved if people were just more honest and transparent from as early as possible in the process, even if it feels uncomfortable. I mean, in sales, you learn pretty quickly. A quick no is a gift. and so if you got on the first screening call with a candidate and they said, Oh, you know what? These three things that you're saying for the role aren't really for me for these reasons. Cool, thank you, or vice versa, Oh my God, how much time have you saved? As you say, they couldn't actually end up in the role itself. and that takes months to then come through, so if you can sort out those mismatches earlier on, just saving everyone so much time and headache and heartache probably

MAX: true

TIM: What about thinking a bit more broadly over the last few years? Are there certain trends that you've seen in the market, like changes in supply and demand, and how has that then impacted hiring?

MAX: That's an interesting question, maybe not that easy to answer. I think that there was a lot of change in the past. I already said before that data is not there since a decade. It's quite new still, so at the very beginning there was a big, big lack of people that do have knowledge in it at all. and for sure did not have long knowledge in those areas, and back then it was very important to get candidates that are super enthusiastic, that want to learn new things that are challenging the status quo, that want to move forward and bring the company to the next level, but meanwhile you can already hire candidates that are out there and that have 10 years of experience and that do know certain things and then can move you to a certain level. So I think this is one change that happened so that in the past you had to really focus on getting people with fire with ideas that want to implement new things, and meanwhile you can balance that and can rather say, Okay, I bring in a couple of people that do have experience, and I bring in a couple of people that want to learn a lot, and together they can really move forward fast. The other part where I think a big change happened, luckily, is that in the past you often looked at the technical skill sets and you said, Okay, is the person good at working with Hadoop or with other tools that were out there and still are? Meanwhile, it's more and more important that the people are more focused on the business value, and this is something that's dramatically changed. In the past I still remember we had projects coming up from the IT saying, Okay, we can now store petabytes of data, and it's all out there; start using it, and then we started looking into it, and we saw, Okay, it stays saved in the wrong format, so we don't get the information itself, so it has to happen the other way around and say, Okay, this is a business value, and this is a technical answer that we can deliver, and the candidates nowadays, you have to prove that there is in the past the focus, the wrong focus, was on the technical part only, saying, Okay, great, you can work with data science algorithms, but that's great, but this alone is not enough.

TIM: Sounds like there's just been a maturing then of the ecosystem in a sense because we're focusing on the stuff that actually matters now in the sense that anyone in any job is hired ultimately to deliver some kind of value, so now we're focusing on that maybe more than we did before.

MAX: And with that, also different roles came right So in the past you rather wanted to have this unicorn that can do everything from importing the data, preparing it, presenting it, and doing some crazy machine learning on top of it, correct? Okay, whereas meanwhile you have so many different roles out there. It starts with data engineering, then analytics engineering, business analytics, and data science. You have machine learning engineers, governance managers, data product managers, and so on. The bigger and more professional the team gets, the more specific roles you need and the more clear you have to be in whom you need in which areas because today, I think unicorns are not really existing, and you really have to cut it down. So that's the thing: the people out there cannot fill the roles right, and it's not just like a dream scenario from a company that you can do everything but rather that you set up the team so that it's possible to achieve the goals from the company.

TIM: I wonder if the closest thing to a unicorn would be like the first data hire in a startup scale-up. What's that almost close-ish? Because they're going to need to be a bit of a generalist, can do a bit of everything, and get the basics off the ground. Have you seen that before?

MAX: Yes, and coming from my past, where I think now it's the fourth time that I'm building up a data team pretty much from scratch, and it always changes, right? It depends on the maturity of the company and especially the maturity of the data department at the beginning. You more need those generalists that are able to do all of those things on their own, that are pragmatic, that can learn fast, and that ship things fast. The more mature it gets, the more people that join the team, the more you can allow to really separate the roles and the responsibilities there, and the more professional it can get, and then at the end you will have professionals for all of those areas that can really boost and bring up the whole team to a new level, which wouldn't have been possible with 10 unicorns in the team.

TIM: Unicorns sound kind of exciting to me. I'd like to see that weird dream team, and what about when you're thinking of hiring into your team? I'd love to hear your thoughts on a trade-off between maybe two things, actually, so one is like their current skill versus their potential, and then also thinking about, like, the soft skills versus technical skills. Like, how do you think about this? What's more important? And to add an extra layer to that, the question isn't already complicated enough. How is Chachapiti changing this? Because it's Chachapiti making, for example, the technical skills a little bit more commoditized, and now you're thinking more about the softer skills and potential, for example.

MAX: Yeah, very, very interesting and good question. Let me try to split it into two parts: so the first one was rather looking for senior people that join right now versus junior people that have quite some fire and attention, right?

TIM: Yeah, effectively. Yeah, exactly, yeah.

MAX: This is something that is very dependent on the maturity of the team itself. When the team is already set up, when you have the data platform in place, when you have the processes, when you have everything documented, when you have senior people in the team that can teach others in how things are working, I think then It's very important to also bring in more junior people, maybe even people from different areas, different directions, that come in with a different mindset, a fresh mindset, that come up with ideas that challenge existing things, that are eager to learn, that can then raise the bar and iterate together with the whole team. When you're a small team, as we had it before, you rather need this unicorn. You rather need this. this senior person that has seen a lot of things already that has seen stage 10 of the company as it needs to be in the future and that can already say Hey, I've made all those mistakes in the past already. So let's not go this path; let's go the other one. So depending on that, I think it's important to make a conscious decision, and if you want to hire for senior talent or if you want to hire for somebody coming in, bringing new ideas and increasing diversity in the team, and coming to your second question, the technical versus the more communication part I personally Yes, when we're looking more into the future, the technical part will become less important in terms of really executing on writing SQL statements and things like that. I think what is and will be very important in the future is that you really adapt to new changes and that you learn new tools, that you're open to that, and that you understand what's out there on the market. And at the end, maybe it is ChatGPT; maybe it's something else where you see, Okay, I don't have to do this task anymore, so I can outsource it to a tool and really deep dive into another technique." Way of working on something like that right when I think about how maybe eight years ago, when you worked with AWS, you had to think about, Okay, how can I bet on easy two instances? You had to put them up from scratch; you had to think about how this really works. Today you have something like Lambda; we'll just click on it, and it's out there. That's it. So the technicality changes a lot, and it's very important that you adapt to that, and coming back to this technical versus communication I think it is very important that you are very good at getting new information, understanding what's happening there, and really understanding if it is valuable for me or not. When this new tool pops up or whatever, translate this into business opportunities, so to me it's more important that you understand the business value that you can also communicate it and discuss with your team what your ideas are. What's new? Thoughts so that you can really improve how you work versus you have technical skills in how to optimize SL statements, which in maybe two years isn't something that you need at all anymore.

TIM: Yeah, and thinking about it now, I wonder whether it's almost technical versus soft maybe isn't the right lens because it's just that the technical skills are going to change themselves because maybe you're not going to write as much SQL or Python from scratch because now you write a prompt to Claude, get that back, and tweak it. and then eventually that's all automated, but you're still going to be using more software because you're going to be using more large language models and interacting more with AI, especially if the ability to create software is now going to be 99 percent cheaper; surely there's going to be more software to solve more human problems. So if anything is still going to be technical, you have to know how these tools work, these emerging tools, but it's just some technical skills now that would probably not be redundant but getting close to that, and we're focusing on higher-value kind of tasks, I guess.

MAX: Yes, definitely, and of course you still need some technical basics, right? But I think what is very important is that you really understand what is happening in the background there. I remember an example from the past where somebody not from the data team took an image recognition algorithm, just tweaked it a little bit, and tried to use it for a specific use case itself and was wondering why it's not working. And when we were looking into that, we saw, okay, the training data is a completely different use case and different things that you had to look out for, so for sure you have to just put your own input into it so that the existing learning, the existing layers, can learn from that and that the outcome will be great. and these are just small things that you need to know. You don't need to write the whole neural network yourself, but you need to understand, okay, what are the things that I can adjust and what are the things that are important that I really have to look into?

TIM: Yeah, I'm so fascinated to see where this will go, and I'm trying to think of analogies from the past of our other technologies that developed, so part of me thinks, Well, everything's going to be simplified; there's going to be this abstraction layer placed on top of everything. You would need far fewer skills. You'd be far less connected to the underlying data science model that you're building; everything would be kind of made foolproof in a way, like I guess a car—you don't need to know how the car works anymore; you just drive it, you know. So I wonder where it's going to go that way, but then surely at some point someone needs to actually know how these things work and needs to be able to scrutinize whether or not we're going down the right path or not.

MAX: Yes, but this will be less and less people out there, so when I'm thinking maybe 10 years back, every data scientist was able to write their own ways of logic and hypothesis and put it into extended algorithms and so on. Today it's rather about selecting the right ones and playing around with that. and it just changes what kind of technological things that you have to do, and it's a couple of companies that really do this professionally and then give you the right interface so that you can work with it, and looking forward, the interface will not be for data scientists anymore. It will be like ChatGPT for every human out there that it's so easy that you can just use it, but that's just iterations that will happen in every area in the future.

TIM: Yeah, what an amazing time to kind of liberate that level of knowledge and skill and computing power to almost anyone anywhere in the world. It's crazy just to imagine what could happen in the next five or ten years.

MAX: Yes, and looking back to what already happened in the last years, it will be super exciting and much faster than we all think in what is possible in the future.

TIM: In your experience in interviewing lots of candidates, there must be common mistakes that they would make along the way in the interviews. I'm wondering if you could share some examples of those kinds of common mistakes that they would make and how they might avoid them.

MAX: I mean, the one part is that there are a lot of candidates that are just not a fit for what the role needs itself. I think there I cannot really give a good insight or tip to the candidates rather than trying to challenge yourself if you're a fit, but I would always encourage trying it out. But in terms of mistakes that you can make during an interview, which you will later regret, to me, they are rather things like not being honest and trying to sell yourself as better than you really are. This, in the best case, results in you getting a higher job title, a higher salary, and more responsibilities than you would get somewhere else. but on the other hand it will also mean that the expectations are far higher and that you have a big, big gap that you can fall into and that you have to pretend to be good at learning on the way instead of saying, Hey, I know those are the gaps; this is what I need to learn, but I'm eager to do it. I'm very self-conscious about it. Take me, bring me in; I will learn fast. You can help me on that, and I promise you in a year I will be there and even better than that, and to me this would be a long-term, far better strategy for every candidate to say Hey, I want to have a company that helps me improve and develop towards the path and goal that I have versus having a company where I need to pretend that I'm a person that I'm not yet.

TIM: So it just comes back to that transparency and honesty again early on in the process. Yeah, I'll share an interesting counterexample to that, which is slightly amusing as well. So I spoke to someone a couple of weeks ago who was talking about one of their hiring fails where they'd hired a candidate and not really validated one of their skills. in this case it was SQL The candidate joined, and on day one they're like, Yep, by the way, I completely bullshitted you; I don't know SQL at all, so there's that. Okay, this is now our shared problem. What are we going to do about that? And obviously the hiring manager was initially quite annoyed and frustrated because they'd been lied to and the candidate misrepresented themselves. but interestingly, to the candidates credit, they very quickly upskilled because they had the ability and willingness to do it, so they basically gained the process a little bit new. Alright, I'm going to bet I believe in myself that I'm going to lie to get the job because I need the job, and then I reckon I can learn this skill within a reasonable amount of time such that at the end of the day everything will be okay. So for that particular candidate it worked out, but for that one success I'm sure there's probably 99 disasters out there where that strategy wouldn't work well at all.

MAX: For sure it depends on how the company reacts to that, and it also depends on how the trust develops based on that, right? So when I get a candidate who says exactly that, I say, Okay, so we wait three months until you can start, and now you tell us on day one that you lied to us, so that's not a good trust base. It takes a lot of time to build up trust, but it only takes a minute or a day to lose the trust, and then you have to build it up again, so depending on how the communication was there, I would not recommend it and rather say, Hey, I bet you that I will learn it, or I would even say, Hey, I will not start in the next three months until I start. I will have learned it; just give me some tips on where I can look into it, and I will learn it on how

TIM: What about from your own experience? Because I was just talking about this other hiring manager, this is like the most memorable hiring fail, but I guess a fail that turned into a success story actually at the end of the day. What about from yourself, either as a hiring manager or as a candidate? Are there any moments that really spring to mind, any particularly good or particularly bad memories?

MAX: Well, one thing coming to my mind, which is maybe a little bit similar to your example before COVID times, back then when it was usual that you invite people to the office, going there to meet there and discuss things, we had a case study. We invited a candidate that came from far away. So he really traveled there; of course, we paid everything for that, and within the first two minutes we found out, okay, he didn't prepare anything for the case itself, so we were quitting it quite fast, and we also found out on the way, okay, he has some friends in the city, so he's also using it to visit them, and I think for him, short time, it was fine. He got some money from us, but of course he wasted time for us, for him, and so on. To me, this was a learning that I said, Okay, before we go into a case study, I want to have all the documents, all the working things beforehand, so that I can check it, so that I can see, Okay, there's a certain level of quality. But what I learned during that time, which I did not foresee, is that this also helps me a lot to prepare for a case study to prepare the right questions to go deeper into that and really have more value in the interview itself, so for me this was a lucky turning point because I improved the way of how I'm hiring and the way of how I'm going into a case study and how prepared I am. So I'm thankful for this person that he did, even though for him I think short term it was also good for him because he saved some money for traveling there, but maybe it was a win-win situation at the end.

TIM: Yeah, I mean, it depends on his personality. If I were in an interview and I realized after five minutes that I knew nothing and I'm getting grilled and these people now hate me, I would find that very confronting. I'm interested in his demeanor; was he kind of surprised that it ended early, or did he know what he was doing? What do you reckon?

MAX: And it's quite some time back already, so from what I remember, he suggested that we could have another interview where he prepares already, and he didn't think that he needs to do anything for it beforehand. So I think for him it was okay, and a little bit maybe of a surprise, maybe even it was a communication issue. It doesn't matter at the end. I learned a lot from that. We have an extra safety net in between, and I have a way to ask better questions and then to

TIM: We've heard a lot in the last few years about diversity, especially within tech, but in all companies, I suppose, and there's been various ways to kind of address this. At the most extreme end of the spectrum, for what I've seen, there's a university in Australia in Queensland that, as of about six months ago, removed merit-based hiring completely from their ethos or whatever their rules are and replaced it purely with positive discrimination. so as in hiring people based on their race, gender, and some other factor, age perhaps, in an effort in their eyes to address some inequities and to have a more diverse work culture, at least that's what they purport to have What are your thoughts on that approach? Is it at odds with merit-based hiring? Do you think there needs to be quotas if we have ways to measure instead? I'd love to get your thoughts in that whole space.

MAX: So first of all, in general, I think from a company's perspective, it is super, super valuable to really invest time into thinking about it and trying to push for more diversity in a team. Just think about it: you have five twins on a team. They all think the same; they will just agree with each other. They will always run in the same direction and will be at the same place where they are today in 10 years because nothing comes from the outside, right? So what you want to achieve as a company is really bringing in diversity, bringing people from different backgrounds. This not only includes backgrounds in terms of what kind of things you studied or what the previous companies were that you worked for, but it also includes backgrounds like how did you grow up and what were the learnings that you had along the way, and you won't be able to have that by only hiring a certain range of the spectrum, right? So you really have to try to foster that and then to make it natural that you're hiring for all kinds of groups that you include all of them, and currently I think from a cultural perspective or the standpoint that we are as a culture itself We need to proactively work for that because when we just leave it as it is, it will stay unequal, and it will fire, as we said before, AI algorithms. They will promote men because obviously it seems like they get the job very often looking in the past, but this is not the thing that we want in the future, right? And so I think it is very good to think about it and how quotas are one example of how we can foster that. I think this loan cannot be the answer; it's not. It gives you an idea to start with or to bring something into your brain to think about, okay, why do we have this? Why do we need this? But a quota cannot be the long-term answer. This can be a starting point that we start thinking about, that we start looking into it and trying to be as equal as possible.

TIM: I'd love to get your thoughts on something. I feel like a lot of the problems with hiring and the way it's traditionally done are because of a lack of measurement and a lack of clear objective criteria at every step where the system is designed such that the best, however you've defined best, candidates are the ones who get hired, and if we just move to an aggressively merit-based approach, that would solve a lot of the diversity issues because there's so much discrimination inherent in the system. For example, the example with the CV applications and just anonymizing it so you don't even have that as a criterion because it doesn't matter; you're just focusing on skills and experience and things that actually matter. If we just had this objective process, wouldn't that get rid of a lot of it? as opposed to saying, No, no, no, the system's broken; let's deliberately be racist and sexist in a different way.

MAX: I think this can only solve it when you are already in an equal world itself, but it does not start with the job application itself that you're currently looking into; it started way before it starts with school times, universities, previous jobs, and so on, where inequality was happening. So if you would put away all the other factors, you would very likely still prefer the people that were preferred in the past, and because of that, have a good CV and a good history, so I think we really are in a stage where we proactively have to push other groups to become part of the team. Right, and when this is solved, then yes, I think this is the solution already: just block the other things out, be open to it, and start hiring.

TIM: and could you ever imagine a scenario where it's like a trade-off? Have you had that before where you're like, Well, objectively, we feel like this candidate is the one again? best quote unquote I'm doing air quotes for anyone listening to this, but they're not the most diverse—again, air quotes—and is that an honest conversation, or are you always trying to hire the best person irrespective of their background?

MAX: It is reality, and I think it's good that we have those kinds of conversations where we talk about, okay, what are both parts? What is the functional fit? What is the equality fit? How fair are we and alone in having discussions about that and thinking about it and saying, Okay, should I hire this person? because this person is much better than the other one, or it should be higher than the other person because actually they are quite similar, and the other person maybe didn't have the chance to get to the same level. Maybe the other person coming from the outside will learn much faster because the other person came to the same level. with different parameters at the beginning as a starting point, right? So I think it's very valuable and needed that we have those discussions at the moment.

TIM: I guess once you start getting into the weeds, then it becomes a question of what the disadvantage is, what's the trade-off between you growing up in a poor neighborhood versus you growing up in an underprivileged ethnicity.

MAX: Yes, and that's our answer. I think you cannot mathematically answer, and you just have to openly discuss it and think about it. and try to find your optimum, which never will be very subjective, and that's a problem we have. I think I'm not sure if AI can help us that maybe getting a bit closer to the real optimum, but the most important thing is that we start thinking about this topic and that we start to try finding the optimum and trying to be fair there.

TIM: Yeah, I'm just imagining now some kind of like privilege metric on a score of zero to one, where one is the most amazingly privileged person ever, and zero is, Oh my god, you had no chance. You need help, and then some. There must be some solid research around, like what impacts your ability to climb through life and collating those kinds of 20 variables. and then coming up with a metric, there you go, and yeah, might be the fair way to do it

MAX: I would love to see that, yes.

TIM: Well, Max, one final question. I'm wondering if there's anyone you'd like to give a shout-out to, like someone who you've learned a lot from in the data space or the hiring space or anyone who just deserves a bit of a high five.

MAX: very good question, and a hard one. There's not something I would say that this is the person that learned me how hiring works or something like that, and now also the kind of person that is observing a lot and trying to get a little bit of an idea here or there, something that is in my mind, which changed quite a bit of how I was hiring as when I had a peer and I interviewed together with him, I saw how he was writing transcripts during the interviews. I did the same, but he just nailed it. He had abbreviations for common things and terms; he had certain topics, and he had a way of communicating and writing it together, and when the interview was done, he made a click, and everything was uploaded, and he was done while I was sitting there 10 or 15 minutes wrapping it up, making my ideas again. Okay, what did I mean with this comment? And so on, so I learned a lot from that. Meanwhile, I have my own logic of doing that and think I'm much more efficient than I was back then.

TIM: Oh, that's a great shout-out and something very specific and a pain point. I've had myself not just in interviews but in sales calls as well; my notes are always a chaotic mess after the fact, and I had to go back and do that rejigging you mentioned, so a good shout-out there. Well, Max, it's been a great conversation. really insightful, really interesting, great to hear your perspective on a wide variety of topics, and thank you so much for joining us today.

MAX: Thanks a lot; it was a pleasure.