Transforming Recruitment with AI: Insights from Timm Rotter

AI is anything than perfect. And you already spoke about this bias topic. I think this topic with hallucinations and deepfakes, it's quite, quite dangerous. And the third risk is one thing I see in many companies that the people do not get the proper training. That is the biggest risk I would say.

Hi and welcome to the show. I'm Johnny Campbell, CEO and co founder of Social Talent. And this week on the podcast we we're speaking with Tim Rother from Disruptive, joining us from Munich today to talk about transform recruitment with AI. Tim, you're very welcome to the podcast. I was wondering if you could give the audience a little introduction to yourself, tell us about your role, your business and why this topic is of particular interest to you.

Thank you Johnny. Thank you for having me and good to be here to speaking about AI. Well, I didn't learn anything with AI in my former life. I come from sports journalism, working as a sports reporter with Bayern Munchen here in Munich. And then in 2010 something really big happened. Facebook came in our lives. The older people in our audience might remember. And then I started a social media agency and that was in a nutshell and we are still existing with the agency, but we saw that technology is disrupting our work faster and faster. And in 2018 we lost one of our biggest clients to an AI. It was a social media monitoring job and the company says hey was good to having worked with you but now we have a tool. The tool is cheaper and the tool is faster. For us it was a backslash for sure. But then I started navigating through this AI world and found something quite interesting growing up there. This was called ChatGPT. ChatGPT is older than most of us know it's not. Did not come up in 2022, but years before. And that's where we dived in in those days started working with that and in last year when everybody started working with AI, we found it Disruptive, our AI consultancy where we help companies work with these tools productively because I think that is one of the biggest challenges now for a lot of corporates, for a lot of companies, after having had two years of happy prompting now to really work productively and effectively with these tools. And that's what we want like to support with our clients here in our Munich office with around about 50 people. And great to being here with you Johnny, because I think it's hopefully a good added value for our companies to understand this big AI impact we have.

So Tim, our audience might be wondering what a former journalist who covered Bayern Munich might be discussing here when it relates to AI, but when we met up in Munich last week, you were sharing with me that the original business that you created back in 2010 around social media, had a large focus on recruiting and supporting recruiting teams. So I wonder if you could tell me about that early history, call it phase one of your entrepreneurial career, just how you were supporting businesses, what type of businesses were you supporting, what were you doing with them? And then maybe we might talk about how that's transformed into disruptive and the AI technology.

You're completely right. And this HR business was among the four first parts we started with in those days, because Facebook, when it came up in the early 2010 years, was a huge opportunity. And I think you know the best with social talent that it's important to being present there to attract people to tell stories. And as I came from journalism, me and my colleagues, we have been storytellers from beginning and how to find good stories for social media, how to prepare good stories for social media, and how to broadcast these stories on social media. That was a challenge where I came from. And I still remember one of my first contracts with one of the big DAX corporations in Germany in 2011. It says one of three posts you publish on Facebook for us has to have an image. One of three posts has to have an image. And where did we come from? That was the time of written texts on Facebook. And now what do we have? We have reels on Instagram, we have live videos on LinkedIn, we have stories on all of these channels. So the technology has evolved so enormously that the way of storytelling has to change too. So it was a transformative business from the very beginning. But what was the base layer of all of that in all these years? Telling good stories because that's what people are looking for. We don't have too few LinkedIn posts, we don't have too few YouTube videos. But what we are lacking in social media is having good stories and telling stories about people. And that is what we what we aim to do. Channels are changing, but storytelling is staying.

And not just the challenge channels. I think when you think about storytelling and recruiting context, most of us think about employer branding and how we tell stories through our EVP and through the different mediums there. I know you've been very early in, as you mentioned in ChatGPT, most people think 2018, I think you made a mistake that that couldn't be right. But it was, you mentioned that was when you first started exploring ChatGPT, because of course Although most of us only know it from the last two years, it has been around for a lot longer. You know, most of us came across 3.5, but there was a 3, there was a 2, there was a 1. But considering your roots in employer branding and that kind of piece, you can you walk us through maybe some of those early use cases that you had around personalization, around employer branding, in terms of what have you seen, what are you working on, what are some of those opportunities just to start off with, when we talk about how transformation is coming into recruitment with AI, let's start with that area that you had so much expertise in.

Yeah, and you mean the time before ChatGPT and the times where.

Well, even now really, Tim, we're looking at, you know, personalized content. So. So maybe anything you can share with us around work you've done to leverage AI to personalize that kind of experience. Because obviously previously we tried to think of Personas, we'd map to five or six Personas. We try and create content for those Personas and then we'd hope that when you see our content, you might resonate with one or two of those different Personas that we're trying to target. That's the old way of doing it, so to speak. Talk to you about what you've seen in terms of targeting folks using AI with very, very personalized Personas. What can you share with us about that kind of an approach to AI empowered employer branding?

If we step, do one step back, what you are already speaking about is personalized context content. That's. I think it's a very important understanding that personalized content is so important because not all companies have understood that, but it needs different content for different Personas. It's not only about the side of the, of the audience, but it's about who is sending the content to. There is the question of personalization too, which is so important that it's not only the brand itself or the company which is publishing anything but that you have very many different channels who might broadcast content. Speaking about corporate ambassadors, speaking about influence and all that. In the old world, before ChatGPT, we had this personalization too already. But now, of course, it's much easier. I think it's much easier speaking about two things first. And then we come to the employer branding part in this long talent acquisition process. From employer branding to sourcing to recruiting to onboarding. In this first part, in the employer branding part, AI has two big levers. The first is that you can easily produce from one basic content very different contents, four different Personas. You maybe have your white paper which is telling interesting stories about the company, and then you want to publish it for different audiences. Of course, that's nothing chatgpt does in a proper way. If you only give it the white paper and then you put your prompt in and say, hey, give me three different versions of this content. But if we use AI in the HR context, I think it's very important that we understand that we have to train these tools in a very individualized way. Otherwise personalization will only lead to mediocrity and not to the kind of content we want to have. This is of course, the content creation part. We can dive into this deeper a little bit later. The second part is something which is not only related to our own company's business, but if we speak about broadcasting this content to these different audiences through LinkedIn, through Instagram, through TikTok, you have a lot of AI in these tools. How they broadcast, these advertisements we produce, that's something very many people do not know. But AI did not only come with ChatGPT in the, in the recruitment process, but is in LinkedIn. The LinkedIn managers, the systems are loaded with AI and they have a very deep understanding about the different target groups. And so we have seen that these kind of social recruitments become more more efficiently through AI and these tools. But if we speak about how AI is enhancing the content creation process and about this Persona concept, what would we do? Let me give you an example. We have a client in the electronics industry. They reach out to different target groups. Recruiting target groups through personalized newsletters in the past was great job for us to write seven different newsletters for different Personas. Speaking about the agency business was quite nice because you made a lot of hours with these newsletters. And this AI is disrupting agency business too, for sure. And now this client is using AI together with us. And what are they doing? They have trained an AI system, an AI chatbot, not ChatGPT, but based on this large language model. And they fed it with seven different Personas and said, hey, for example, you have Johnny, the classical entrepreneur guy, he's 22, 25, he came from university, he's in university, did his masters in this class, for example. And let's write an internality which is, which is a good fit for him. And you have the seven other personalities in this system. And putting in one content leads to seven different kinds of contents which are produced automatically, but only because the system is fat before in a proper way. And it's not perfect. It's really good. We learned that we save about 75% of time through this automation process.

Wow. So significant savings, I imagine, for a fraction of the cost. Like, I doubt you're, you know, spending three quarters of a salary of somebody to do that writing. So not just time saving, but an enormous amount of cost saving. Would that be fair, Tim?

It is, it is time saving. It's cost saving. And on the other hand, of course, we have to understand where AI comes to its limits because in the end, we as the humans, we have the final decision if it's proper content and even the best AI systems. And I think that it's very important speaking about it in such a podcast, because it's not about praising all these new technology which is coming up. In the end, we have the necessity and we have the duty to have a look on that because even the best AI systems are still hallucinating. And sometimes you have two or three great. We had this in this project, two or three great paragraphs of the newsletter text and in the fourth there was a hallucination that was not understandable. If we had sent that out, everybody would have seen that it's not human quality assured content, it's us who stand for quality.

So it's still, as you would say, approved by, reviewed by a human and that's still necessary. What you've seen on scale is that you still can't just turn it on and let it send that stuff. You do need to keep a human in the loop, so to speak, to approve this.

Sorry.

So we're going to think, like from employer branding and attraction perspective, we see that scale, personalization saving 75% of the time. Let's go all the way to the other end of the spectrum, Tim. We've made a higher. I know you've used a, you've been, you've, you've advised a lot of customers on the use of AI and technology, video technology in onboarding. Could you maybe talk us through some of the work you, you've helped customers with there? What's, what have been some of the ideas, what are some of the savings and benefits there?

Now we're speaking about the situation when we already made a hire. Right, Right. Well, Sam Altman said something quite interesting. He said through AI, jobs are definitely coming, going to go away full stop. And he's right. Because in all these steps, a lot of jobs are changing and a lot of jobs we are losing. Nevertheless, new jobs come up and in the end, it's not AI that is going to steal our jobs. It's the AI users who are going to steal the jobs of those who don't use these tools. And I think the situation you are describing is quite a good one. In the past, when we made a hire, we had this long lasting onboarding process. You either prepared your new candidates with dozens of pages of PDF files to understand all these rules, they had to know in the company how to find the canteen, how to behave according to the GDPI and so on. And maybe somebody read these hundred pages, most of them won't have read that. The more modern companies, what did they do? They produced nice videos which explained all these rules you had to follow. And unfortunately then paragraph seven of the GDPR changed and they had to invite the video team to come in again and produce the entire video in a new version for another 10,000 Euro. And now the world is changed by AI and we are working in projects like that with a tool called heygen. Heygen is a tool which produces AI avatars. And these AI avatars are produced once and then can say everything in all languages you want to let them speak. And you can easily prompt in the back end the text, like in a Word document. And then the language is appearing. And my avatar is so good that people really do not understand. If it's me Orbital avatar here, it's me, Plonny, no worries. But it's really difficult to understand that. We had a nice case half a year ago when our head of video came to me in the office and asked, hey Tim, I would like to try something with AI. Would you be so kind to come in the video studio? I want to do a short broadcast. She broadcasted 30 seconds with me, asked me something about my life and then said, okay, thank you, I'll come back to you later. Half an hour later on our intranet, there came up a video clip from me where I said, hey guys, you did such a great job in the last days. All of you have a free day tomorrow. Of course that was not me, that was my avatar. But the quality was so good that really people called easy how I had on video. Is it right? Did Tim really give us a free day for tomorrow? And then Izzy had to write in the Internet. No, sorry, that was a deep fake. Unfortunately we have to work tomorrow. And the quality is so good that even people that know you do not understand if it's avatar or not. And this was funny stuff. But let's think about the onboarding process. We have a client in the insurance world, they have to produce all the videos in 17 languages and Dominic, the colleague who's there in the HR team is using the same tool and Iki head has to do an update for the onboarding process and he uses his avatar and gives advice to all the people in all their mother tongue. And then the colleague from Paris last time called him and said Dominic, I never knew that you should speak French so perfectly. Why don't we speak French? And said Dominic, sorry, that wasn't me. And it makes it much faster, it makes it much cheaper and the quality is almost on the quality level we have with real video and it's not in the same quality but if you understand, if you let the people know that's avatar based videos, the acceptance is already very high.

So far Tim, we've talked about how the recruiting team can be using AI is very powerful. But let's not forget candidates are potentially using AI today, for example, it will share with me a list of tools from final round. AI offer Genie interview copilot Lofa Verve copilot interview with AI locked in AI aspect HQ AI pure job bump offer in AI creati AI interviews chat comraku ke offer goose.com this endless list of tools that I was quite fascinated by this Tim, but they are tools that can of course allow a candidate to write a resume or write a cover letter, apply to a thousand jobs at once. But also they give them in screen apps that work in Team Zoom. Any interview client that show them live what to say, how to respond, guide them in the interview process, like really sophisticated real time recommendations on what to say and how to respond to a live interviewer and the interviewer doesn't have a clue. This person has all these tools available in their window, like the equivalent of having those cool goggles you see in sci fi videos where all the information you need is coming up in front of you. The person you're talking to doesn't even know like what do we do in that kind of a world? How do we navigate that as employers when candidates have the same tech or better tech available to them to get a job.

Yeah, yeah. I think navigating this AI candidate world is a difficult challenge. I like these posts on LinkedIn where they say the 227 best tools in the AI world to do this or that, I think that's, that's shitty content. It does not make sense to think about 227 tools because if you want to work professionally with tools, you have really to deep dive into them. And we in our AI guidance here in the company, they say we don't use more than 10 tools because you want to be experts with that. That's to this, to your list you spoke about. But what's happening with the candidate in the candidate world? Yes, we see that too and we learn and when we discuss it with clients, it's becoming more and more difficult to understand with whom we are in exchange. Is it the real candidate or is it two people? Is it the candidate plus his digital body he has there? Yes, it is the digital body. And to be honest, I think we should be happy when our candidates know how to use these tools in a. In a proper way. Because it is not a specialist technique to prompt properly with ChatGPT, but it's a culture technique. Everybody has to learn. And the better the people are, the more future ready they might be for some jobs that come up in the, in the. In the future. Nevertheless, how do we, how do you understand what's real? Spoke with a, with a recruiter colleague. They have the same challenge and they say all these written pre assessments we did in the past and even the, the. The virtual interviews will lose importance in the process. In particular, if you are speaking about high level positions, then it's the exchange, the personal exchange in real life which will become more important in the future. I think it's funny we have more technology, but the analog exchange becomes more important in some situations. On the other hand, yes, the tools are better and you have some real time in app tools in your video conferences. But you only can use AI on a professional level and in a topic, if you're really an expert in that. If you started to interview me now about, let's say semiconductor production, even the best in app tool would not help me because I wouldn't do it so fluently, I wouldn't do it so convincingly. And then you, you change from. In. From semiconductor production to. To something else. Let's say in my last holidays in Paris, AI is not so authentic, so human. The recruiters will know that and in the end, Johnny, it's your business with social talents to train the recruiters. They have to be retrained too. In a way. They have to understand more the human touch and the human touch will become more important. Nevertheless, I'm happy for every candidate who is an AI pro because we all have to understand this big technological disruption we are just experiencing.

I like your point. Reminded me a bit of the French play which became a couple of movies. Serrano de Bergerac, when Roxanne was the movie that Steve Martin made. Famous in the late 80s. And we filmed with Gerard Depardieu as Serrano de Bergerac and Serrano later on. But point being, it's a play from the 1700s where. Where the person was whispering beside the. The ugly guy was whispering to the person who was handsome, what to say to the beautiful woman to make her fall in love. And eventually, of course, the sham was revealed and the. The. The person who was, you know, not so beautiful, but had the beautiful heart and mind was shown. And the person who was just saying all the right things but didn't actually have the substance, you know, what. What didn't become the hero. I think you look at that and you think about AI and candidates and interviewing. It's. How do we uncover the Serranos, you know, in those situations? And I think you're right. It comes down to the abilities of the interviewers. I think we have to be much better at assessment. We have to know our domain better. We have to be able to dig in and probe into examples. We have to be able to figure out, does that sound real or not? Does that. Is the context correct? We have to be able to pivot those questions. So we have to be quite confident in our interviewing and assessment skills as recruiters. But being honest with you, Tim, most people like me are not. They do not have that competency. They aren't strong there. And you're right, it's a new set of skills. Maybe they've been focusing on how do I source talent, how do I come up with candidate Personas which they don't need to do anymore. So I think you do have to move your focus as a recruiter and think about these things, because the candidates are going to continue to use the technology and. And it's going to get better and better. So we have to keep up. But when I think about this, you know, I do admire a candidate who has invested in their AI skills and have found one of these tools, set it up, and mastered it. And I kind of think that's cool. I'd like. I don't want them to cheat. I don't want them to misrepresent themselves. But I do like that they're experimenting with AI and using AI tools. I'm going to want that in the organization. Do you think they're also looking at the same. Looking at organizations the same way to see how open is this company to using AI? Are they technologically savvy? If I go into this company, will I be able to use AI to do work? Will they support me? Will I learn will I get access to great tools. Is that becoming maybe a differentiator for organizations in attracting talent, how they use AI themselves?

I'm sure it is. I'm sure it is. And it will stay for the next years. I think in some years then it won't lose upon a bit because then it's natural everybody will have started using AI. But right now it's a real differentiator and we learn in our agency and I said we are quite a small company with about 50 people working here. We learned that it's a differentiator. We won one great colleague last year for instance, she came from a very big agency and she said and no, maybe it's not so glamorous job when I work with you here and maybe the other company will have the bigger clients. But you are, you are really in the forefront of innovation in AI. I want to stay to be in your team and I want to be there. And it was known the only one and others applied here and said hey it's cool to being in a company which is, which is really upfront. And on the other hand I have a good colleague working for a big pharma company here in Munich. And what does this pharma company do? In the very beginning of this gen revolution they forbid to use all AI tools and said this is dangerous, we don't want you to do that. And they did not change their mind for a couple of months and they really lost talent. My colleague left the job. Another friend of him left the job too because I said I don't want to work in a company which is not future oriented. They have to give us a picture how to involve this technology in our, in our daily business. It is important to win attractive talent to be, yeah, as you said, future savvy and able to give people a big picture how to work with AI professionally.

And I guess it's going to start in the hiring process. If you can see that an organization is actively using AI in the hiring process, encouraging it in a responsible way. It says to you this organization believes in this technology, invests in this technology. I was only thinking last night one of my kids, I have four boys and one of my 10 year olds we talked to, we've got twins. I have twins. You've twins as well. My one of my 10 year old boys had to complete a project on Japan for school and his teacher told him to go home and search on the Internet about Japan and he said dad, can I, you know, use your laptop and Google search? I Said, well, Google search for this. No, that's like two, that's two years ago. I said let's just ask Chat tpt. And I, you know, I showed him how to prompt and you turned on voice and he spent an hour and a half I listened into him and he was prompting chatgpt brilliantly tweaking his answers. Make that shorter, make that whatever, give me different information on that. Looking up manga, looking up sumo wrestling, all these things and learning, like genuinely learning. Somebody else might look at that and say, oh, he's cheating because he's asking cheap chatgpt. No, no, he's searching for information, he's putting it together, he's augmenting it. Then he's handwriting with a pencil into a journal article about Japan. And, and, and he's learning, I think about that. And, and, and kids like that who will go into the workforce. We want those kids to be brilliant at those tools to get work done when they join the workforce. We want colleagues of all ages to be brilliant at these tools. But are there risks, Tim? You know, like, you and I are both very, very positive about the benefits of AI in recruiting and hiring and in organizations in general. More are investing in this, in our businesses. You know, we're on the bus, we're supporters. But should we be sitting back and looking at any genuine risks like bias and everything else? What's your stand on this? And you know, for sure, can you share, can you share your opinion?

Yeah, for sure, for sure. There is and I quite think it's quite important and to be very, very serious about that and because AI is anything then perfect. And you already spoke about this bias topic. I think this topic with hallucinations and deep fakes is it's quite, quite dangerous. And the third risk is one thing I see in many companies that the people do not get the proper training. That is the biggest risk I would say right now. Because if people are not trained properly in organization and they start doing it on their own, start using tools which maybe are not approved and they are bringing their own devices and starting with that and they put data in these tools. They should not. So first things first, proper AI training in all organization should be mandatory. Otherwise people don't understand the risks and only take this data security risk. Anything you put in an open version of ChatGPT will be stored on OS service. And deep SEQ does not make it better. Deep SEQ does not use US service, but they put it on the Chinese ones. And speaking about gdpr, that are all infringements which happen immediately if you use these tools. So data privacy is the first thing. Second thing is biases. That is quite important and not so many people know about that. And in particular in the HR process, of course we say technology does reduce biases because the human biases do not come up then. But that's not true. The tools are as biased as we are. Maybe you know this Amazon fail a couple of years ago. Their AI driven hiring model was a little bit groundbreaking, but it does not work because in all the hiring processes it preferred male candidates for technical roles. Why did we have this prejudice? Because it emerged from the, from the scars of historical gender imbalances within the company. And therefore the tool thought, okay, we hired male candidates for tech jobs, so the next one which should come in our team must be a man too. And so the female person did not have the same opportunities. We saw it here in Munich when, when a news channel unveiled that an HR recruiting tool that broadcasted candidate interviews and gave recommendations about their characteristic and their capabilities was not fair, but biased. What happened? This tool judged differently about one and the same person depending on if it had a scarf above his head or not. And the same person wearing the scarf did get worse notes from the tool than without. So there are a lot of biases in these tools. We do understand. We do not see because the tools are biased in the same way we are biased because we trained the tools with our data. And take these Amazon, these poor Amazon engineers a couple of years ago tool, they put all these data in and they knew, okay, technical candidates are men. So they did not understand that the tool was biased because it gave them the same output they had learned, which was right in the past too. So this bias problem is the second big thing. The first thing is of course hallucinations and deep fakes. These tools make mistakes and even the best ones make mistakes. And that's why we have to stay in the driver's seat. Otherwise we cannot be sure that it's really safe content. And if you have the word min machine interaction, there's a reason why min stint in the beginning, because we are the ones who have to be the deciding choices in the end.

Tim, you're working with many organizations, helping them embrace AI, train up on AI, run workshops, become fluent, conversant, look at different use cases. If anybody listening to the podcast today is interested in hearing more about what disruptive does and engaging you and your firm, or joining some of your free webinars, or reading some of your articles, what's the best way to contact you and find out more about you and your business and the thought leadership you put out there.

First of all, thanks for thanks for the stage, Johnny. While I'm on LinkedIn, Timotha for sure. So if anybody has a question, let's go in exchange there. And besides, we set up an AI academy, a digital one for companies to help them that their people can learn AI according to their personalized learning path. And the URL is disruptive minus academie, the German spelling.de but you find it with disruptive and AI agency.

Yeah.

So thanks a lot for having me. Johnny was a great exchange. Learned a lot about your position on these tools and AI. Stay in touch and everybody out there. One message to the end to everybody out there, AI might not be perfect, but the best time to start with this is now because the tools are such a big lever in the recruiting world. So let's get started and try it out. Thanks for joining, Tim.

Thank you so much for joining, sharing your insights, sharing those stories. We're going to put links to your academy in the show notes and to your LinkedIn profile. That's Tim with two M's, Roger with two T's. Real pleasure to meet with you last week in Munich to hear more about your business and to have you call on the podcast this week and share those insights and stories. Hopefully that's inspired some other folks here to think about some of the opportunities, use, cases, risks. But as you say, there's no better time than now. Best time was yesterday. Best time, second best time is today. Wait till tomorrow. Jim, thanks for joining the show and thank you, the listener, for joining us this week. We'll be back with another podcast this time next week. Don't forget to subscribe wherever you get your podcasts on and we'll see you soon for hiring excellence.