Highered Podcast Ep.1 | AI In Recruitment and Talent Attraction (Christian Halvorsen, Vend)

Yes, thank you. Happy to be here and talk about some interesting topics with you today. So as you said, yes, I'm CEO of Vend, a marketplace company working in the Nordics within four verticals or categories. It's mobility, real estate, re-commerce, and jobs. And that's what we're here to talk about today. And I have quite a long history of working in marketplaces.

Absolutely, I think it's quite interesting because as I said we operate within these four verticals and we apply AI to all of them. But what I notice very clearly is that the jobs and recruitment is the area where it's let's say most unstructured data.

In the form of texts, role descriptions, advertisements, CVs, application letters and so on, while when you are in mobility for example you have a car it has certain very strict features, horsepower and size and so on, but in jobs it's more unstructured and that's actually what AI is very good at processing, right? So I actually think that AI is and we see that within our services this is where we use AI or come the furthest in using AI just because of that.

No, I think it is actually that's when you can use AI to draw out information that is otherwise harder to do.

Yeah. And especially then and if we talk about the direction I think especially when we talk in scale and in volumes and I think that's one of the concerns when we're talking for example with recruiters maybe that they will see a surge in number of applications, how they are written, that they improve dramatically because the research requirement to actually apply for a job and find a job is dramatically going down.

Yeah, I think so. It's a little bit hard from our point of view to know the numbers of it, but I'm pretty sure that that's what we are seeing. Yeah. And do you see that how far have the in your experience the companies or the ones who are actually looking for talent come in actually handling these increased volumes? How are they responding? Have they started to respond?

I think it varies and I mean we focus on a relatively narrow part of the process. I told you this before the recording that we focus on the advertisement part of the process, not so much the selection process as such, but from the dialogue that we have with our customers we see that recruiters are starting to use these technologies but to some varying degree of course depending on the company.

It's a little bit hard to say whether there should be concerns. I mean there are new technologies of course then you don't know exactly how it's going to play out but I have a generally positive and open mindset to this. And I think that probably the upsides and benefits to using AI in recruitment processes outweigh the downsides if you use it in a responsible way. And that is of course up to players like us and also all the recruiters out there because one concern could be that if the volume of applications increase dramatically then you would need to rely more heavily on some kind of filters or help also to screen this.

Well, I think to some degree you're right. These AI systems are black boxes in a way. But I think that when they are used to for example screen or filter based on certain criteria I think they can be quite good also at saying why they did certain things. And you also asked about bias. That is of course an underlying concern and should be a concern but we should also remember that human processes also have bias so we're not comparing with a perfect process today and in fact I think that the AI can actually help reduce bias because you can actually tell it to be aware of it and make you aware of when there is bias in the process. We can just give one example from our own services. We have actually now created a tool so that when you insert an ad onto our system, it will point out where you have non-inclusive or non-diverse language in the ad and tell you to correct it to make it more inclusive and attract a more diverse candidate base. So I think it can actually be used to help out on those areas.

I think it's a very interesting example and I think that AI can be quite good at picking out candidates that match certain requirements and so on. What is probably harder is that there could be really good candidates that don't fully fit with the requirements but they are still outstanding in some way that only an experienced recruiter can pick out but who knows in the future what they can do. Yeah, it's so interesting part also then but coming back to the matching part in terms of what kind of job ads and this is I assume you have also technology for example for showing deciding what car to show me and what job ads and it's important part of your business. How far would you say the matching when you how do you develop a profile for example for somebody coming in and looking for a new job and do you to what degree do you leverage AI in that structure versus your own algorithms that you built over time?

Exactly, that's a great question and it's a topic that we have worked quite a lot on for a long time. I mean over time we've had of course regular search. We've had recommendation algorithms that use your previous browsing and search history and so on to pick out the right ads to match you with. But we have more and more now used AI for this. And we're doing two things. One is to create better profiles on the candidates. Some of that is coming from the candidate themselves creating a profile on our site and we have actually found that you don't need a full CV, you can have a quite limited profile of what you have just done and what you're interested in and so on and that is quite useful and we can actually also enrich that profile based on what you have been looking at, what kind of job opportunities and so. So that's one side of it, part of the matching to have a good profile on the candidates. The other thing is that we have started to use natural language search instead of the old filter based search. So because it is quite hard for candidates to find the right opportunities, but here you can actually chat with our service and just tell them with natural language what kind of opportunity you're looking for and then you get a result that matches with that and then you can continue to say well a little bit more of that than it should be in this region and so on and that is proving to be quite useful and we see that those who use this we call it smart search they tend to convert a lot better to actually sending an application than those who use the old traditional search.

So I would say we're using it for on both sides. So it is both to create the profiles for the candidates but we are also using AI for the recruiters to help them create good ads, good role descriptions, and when they do we help them write it with inclusive language and so on but also based on what they have written we can improve the style of the role description and extract what are the key requirements for the job that helps with the matching. And then I think it's also when it comes to the matching I think the way we think about it is that it's both about matching active candidates those who go in and actively search for a role but also the passive candidates those who are not necessarily searching for a job but they could still be relevant for an opportunity and there we of course use what we know about them and on our services. We are lucky to have all the categories as well. So you could come in and look for a t-shirt but you're exposed to a relevant job opportunity, right? Which is one way of reaching passive candidates.

So we are doing both. We are both kind of doing it on our own site but we are also using the same profiles to attract candidates on social media or news media and so on.

So it's useful for both.

One thing that I do see is that of course if you look at the EU it's quite a bit stricter on the regulation than other parts of the world. So in the EU you have the AI Act which is actually coming into play when it comes to recruitment because recruitment is considered high-risk activity. So in the EU there will be regulations related to AI usage within the recruitment process. So I think that is maybe one area that is different from if you compare to the US for example.

It is coming into effect next year.

Yeah. And now the regulators are figuring out how which parts of the recruitment process will be regarded as the high risk and which will then have certain compliance criteria and so on and which will not be that.

Absolutely. And I would just encourage people to start experimenting and testing AI tools. They will not be perfect but they will never be as poor as today because they will only improve going forward.

Thank you.