EP135: Alooba - How to Navigate the Surge in AI Job Demand

All-In Recruitment is a podcast by Manatal focusing on all things related to the recruitment industry’s missions and trends. Join us in our weekly conversations with leaders in the recruitment space and learn their best practices to transform the way you hire.

This transcript has been edited for clarity.

‍

‍

‍

Lydia: Welcome to the All In Recruitment podcast by Manatal, where we explore best practices, learnings, and trends with leaders in the recruitment space. If you like our content, please subscribe to our channels on YouTube and Spotify to stay tuned for our weekly episodes.

I'm your host, Lydia, and with us this week on this special edition is Tim Freestone, founder of Alooba. Good to have you with us, Tim.

Tim: It's a pleasure to be here, Lydia. Thank you so much for the invitation.

The Hiring Pain Point That Sparked Alooba

Lydia: What inspired you to create Alooba, Tim? I mean, what gaps were you aiming to address with this platform?

Tim: So the story starts about six years ago. Alooba is six years old, almost to the day, and I used to work in the travel industry for one of the major players there, leading the analytics function of this business. A big part of my role was to hire analysts, data scientists, SQL developers, those kinds of characters. We took a pretty typical approach, as any company would. We put up job ads on LinkedIn, manually screened CVs, did the phone screening interviews, and did our own take-home tests. There was a lot of work.

My biggest pain point was always interviewing candidates in that first round who, on paper, had the right skills and the right experience. Their resume stood out. Then many of them, in that first interview, showed they didn't have the skills they said they had. So, I realized how inaccurate a tool a resume was when deciding who to bring into that interview round. We had these take-home tests that we gave candidates, where we gave them a real dataset from the business. They would go and do some analysis, do some modeling, answer some different questions, come back, and present it.

That was really effective, but it was a lot of work for candidates to do, especially early on in the process. It was a lot of work for us to grade, so what I was looking for was some kind of simplified version of that that would be online, that we could give to all the candidates who applied, rather than just manually reading the CVs and trying to come up with a more effective, more accurate way to screen candidates. And so that was originally why we created Alooba.

Lydia: Now, you've already built, and people are using this, and it's grown in the past six years. So what's the most surprising thing that you've seen about how companies approach the hiring process and how they use it?

Tim: For the first five years of the business, we were focused on data-related skills, and so we would typically be pitching our product to leaders of data science teams, data analytics teams, BI teams, and so they'd be the main users of the product. An assumption I always had was that they would be quite data-driven in the way they think about hiring because in their day-to-day jobs, they are running marketing analytics, product analytics, and data science teams, like it's all about making data-driven decisions.

So I thought that would translate across to people and how they hire people, but it doesn't at all. Most people, even in data professions, still make the kind of gut-feel-based decisions, which I find really surprising.

Lydia: So they have the data on hand, but they still go with instinct? Is that what you are saying?

Tim: At the end of the day, no matter if a data leader or anyone had a report that said, objectively this is the best candidate for the role, and they thought otherwise because they would meet them and they just did not like them, they were never going to hire them.

So no metrics and no data are ever gonna trump just their core intuitive belief. I would not be surprised if that would apply to anyone. But, to a data leader whose whole job is around using data, I was kind of surprised at how dismissive they would be of the use of data in hiring to make people decisions.

Lydia: So, how do you circumvent that? Is there an element of counsel in this?

Tim: I think lots of things have happened through time if I think back to the last six years. One is just the fact that we have not tried to oversell our product in terms of what it does. So, we would say that it's a bit of a hiring puzzle. It is not going to give you the complete picture of everything you need to know about the candidate because, ultimately, they are the ones who do the interview. Even if it is kind of subjective, vague, and based on their gut feel, they are going to do it anyway.

So we came in and said, “Well, okay, instead of having to create your own test, which is a lot of effort, why not use our product off the shelf? It's already got all the content. It is an online platform; it has got cheating prevention.” And so we ended up kind of just repositioning our product a little bit in their eyes.

Can Traditional Recruitment Keep Up with AI?

Lydia: So now we are going into a different topic, which is a wider one. We are seeing exponential growth and increasing demand for AI-related jobs. Our team at Manatal has actually analyzed more than 100,000 AI-specific job listings across all the top job boards just to see if there is a demand for AI talent, and it has peaked.

The data that we found has shown a staggering 373.6% surge in demand for AI-focused positions as recently as the last quarter of 2024 compared to the previous year. The question is, are recruiters and agencies equipped to source and assess AI skill talent on this kind of scale and this increase in demand?

Tim: I do not think so. No is the very short answer, and I think there is a fundamental challenge in the way recruitment has been set up in the sense that a recruiter, either in an agency or in a talent acquisition team, is normally a generalist. For technical roles, it could be an AI role, it could be a software engineering role, or it could be a data role. They themselves do not necessarily have any skills or background in those roles they're hiring for. Yet, they are being asked to do a lot of those initial screening calls and the searching and the sourcing, which I think puts them in a very challenging position.

I try to empathize with them and think, if I tried to hire a civil engineer or someone that is like completely outside my skillset, honestly, I would have no idea what I am doing. At best, I would be able to ask my friend who is a civil engineer and get some feedback from them. Maybe after I work in the industry for a little while, I will start to get a sense of the lingo, but at the end of the day, I am still not a civil engineer, and I cannot really get to the bottom of their skillset.

So I think that is the same with any kind of technical role, especially with AI, because it is changing so quickly that even an AI expert really does not even know necessarily what the right skills are because there is a new model released every week.

So I think it is a very difficult challenge. If I were a recruiter or a talent team, I would be looking at involving some kind of platform that at least gives you a sense of the skills of a candidate. That would be one way to go about it or relying more heavily on the hiring managers in the business who are experts in those fields.

If I think back to lots of the customers we spoke to over the years, normally analytics leaders, they would have ended up taking up a lot of the burden. I got a sense that a lot of the time, they did not quite fully trust their recruiters to do the screening step for them. Again, it is because they are not an expert.

So they would say, “Hey, you send me the resume and don't worry, I'll screen them.” I suspect we will just see more of that, and it is a difficult challenge for recruiters to solve.

Lydia: What might be some immediate steps to bridge that gap in knowledge and expertise?

Tim: I would be thinking of either engaging some kind of expert in the process, and maybe recruitment agencies should be thinking more of domain experts as recruiters. Is it beyond the realm of possibility that you hire a former analyst who is an analyst consultant and get them in as a recruiter because then they add that domain knowledge and expertise? I have seen a few recruiters in Sydney take that approach. So that will be one option.

Another one is to use a third-party assessment tool of some description to then give you some kind of a sense of the person’s skill. A lot of the time, I think they will end up relying on their client where they present their candidates do the majority of the screening on their side. Sometimes, the clients are happy to do that because they realize and expect that the responsibility is on them to do those technical evaluations.

Untapped AI Potential Across Industries

Lydia: Going back to that surge of 300 percent that we are seeing in AI-related jobs, right? Will this huge growth sustain itself, or are we witnessing a sort of temporary bubble?

Tim: I think we have not really touched the sides yet of what these large language models can do. One thing I like to think about is, even in day-to-day life, all the problems that we have and all the ways that AI could improve that is you might be able to see behind me, even just the plants back there, which are real, by the way. One of them was looking a little bit sick about a week ago, and I was removing it from its soil, and it had some weird, funky-looking stuff in there. I used the video mode of ChatGPT. It gave me an immediate analysis of what was wrong. It told me what to do with it. And so that is just one tiny problem. You can imagine thousands of other similar things that we are not even really thinking about.

Even the most avid AI lover is probably not using it at its fullest at the moment. When it comes to business and hiring in particular, we are not even touching the sides yet. If I think of all the people I have interviewed on our podcast over the past four months about how they are using AI around hiring as an example, we are not even touching the sides.

Maybe help with writing a job description, maybe summarizing some interviews. But in terms of all of hiring, could it be done with AI completely? And so there have got to be so many other domains in business where we have not even touched the sides. And that is just in certain industries. There are other industries that do not even use spreadsheets to make decisions yet, that are not even a kind of ground-up when it comes to using data.

And so they also have a long way to go. I think there is so much upside, and I do not think we have [Unclear] the growth of AI yet at all.

Lydia: Definitely. So, we are going to see more jobs coming out evolving from the jobs that we're seeing right now, right?

Tim: I think so. And it is also probably not just AI specialist roles that have clearly surged, but also expectations that anyone in any role would have some kind of AI knowledge. Obviously, it's not at the same level. We are not building core models, but there is an expectation that you would know how to prompt ChatGPT to get a reasonable answer.

Maybe you have used another one of the models and just almost like that acceptance of using AI in day-to-day jobs. I think that is where a lot of the growth in demand for the skill set will come, as opposed to AI specialists.

Lydia: And so, Tim, on that note, what industries or roles do you think are most affected by the rise in AI-related jobs these days?

Tim: I think so far the best use of large language models that I have seen has been in content generation and coding. I feel like content marketers and software engineers are probably the two areas where I would be looking very closely and keeping up to date with how they go and using them to your advantage.

In the example of a software engineer, I think it will get to the point in the next year where the idea of writing code from scratch yourself will become very silly because it is very inefficient, and it will be more of a model where you tell the AI exactly what you want, it produces the code, you QA, and it is done.

And so I think there is going to be an adaption needed from people in software engineering roles to rethink their core function—almost like a repositioning of them as more like a product owner rather than a coder. I feel like that is a big one.

I think written content as well, from some of the stuff we have done at Alooba, it is staggering the volume and speed at which you can create good enough content for content marketers. I would certainly be keeping an eye on that.

ChatGPT, Auto-Apply and the New Job Hunt

Lydia: Now, we will move on to job market dynamics. We are seeing figures from LinkedIn. LinkedIn is reporting over 220 million people globally have listed themselves as open to work, and that is a 35 percent increase from last year.

Is it really getting harder for candidates to land a job these days? What are your thoughts about using that badge?

Tim: Yes, I feel like there is a market downturn, so that is one major factor in why it is more difficult to find a role—because there are fewer jobs and more unemployment. I would not be surprised if we went back five years to the last downturn. There might have been a similar peak in those Open to Work badges. There are other factors at play.

One is the fact that candidates can now use ChatGPT to generate a CV and automatically apply. We were even playing around recently with some browser tools that just go and apply one by one to every job for you on LinkedIn, Seek, and Indeed. I think that probably each person out of work is applying to way more roles than they were before.

Then the other thing is there is more transparency over applications per job than there used to be. I know on LinkedIn you can see the application count. I do not think you could four or five years ago. I wonder if that transparency feeds into the narrative of, “Oh my God, I cannot get a job, I have to compete with a thousand people.” And so maybe that is also creating a sense of a bigger problem than what there is in reality is what I’m trying to say.

Lydia: So with assessments being one of those stages that you have to go through before you can get an in-person interview, do you, from your experience, see an increase or decrease in the number of people who are underperforming in assessments?

Tim: No is the short answer. So we haven't seen great inflation or deflation through time either as a result of people using AI tools or people being out of the market. So no, we haven't seen much change there at all so far.

Lydia: Now, going back to the Open to Work badge, there are some people who view that negatively, while others say it is a way for candidates to get all the help that they can get. So, what might be your perspective on this divide?

Tim: Yes, I think if I were a candidate looking for a role and I was out of work, I probably would lean against using it because I think the downside is probably greater than the upside. Even if we are having this debate, I feel like that tells us what the answer is. Because let us say it is fifty-fifty—half of the people think it is desperate, half of the people think it is a good idea. Well, if half of your audience perceives it as desperate, then you have to go along with that. So, I probably would not use it if I were looking for a job.

I would take a very strategic approach. I would focus on leveraging my network, trying to get a foot in the door behind the scenes. Who do I know at a company that I want to work for? Or my top ten contacts—who do they know that is currently hiring that might cross over with what I know? I would work behind the scenes to get some kind of warm coffee introductions and those kinds of things, rather than applying with everyone else because it is very difficult for anyone to stand out when you are applying with another thousand applicants, and you could just be another resume.

So I would try to go behind the scenes, and I probably would lean against using the Open to Work banner myself.

Lydia: It is building value through your network.

Tim: Yes, this is admittedly a long-term play. I think if you are an early-stage career graduate, you probably do not have much of a network established. And if you are suddenly out of a job, you probably do not have weeks or months to have these coffees. But given enough time, I would definitely take that strategy.

Networking is a long-run game. You might not get anything out of it today, but if you plant these seeds, they will grow, and you can help each other through your career. I can think of numerous cases where that has happened to me personally. Even the investors in our business, from whom we raised capital five years ago, were people I used to work for.

And so that is a great example of turning up on time every day, doing a good job, and networking transforming into something really powerful.

AI vs. Integrity: Rethinking Candidate Testing

Lydia: Tim, let us talk about assessment—the integrity of assessments in an AI world. What sort of innovative assessment approaches have you found to be most effective when it comes to evaluating real skills while minimizing, as you mentioned earlier, AI-assisted cheating?

Tim: It is tricky. Whether using AI is considered cheating is a gray area. Would we want to discourage someone from using the tool that we would want them to use on the job? Versus the idea of saying, yes, that is great, but I want to get a sense of the raw candidate and add AI to them later.

Because of this blend of opinions, the way we set it up at Alooba during recruitment is that they can either turn on or off various features to make it a more or less aggressive cheating prevention setup.

Lydia: How does that work?

Tim: So basically, you could say, “Okay, I want to have screen recording on. When the candidate starts a test, they have to limit themselves to one monitor. The company can do a full-screen recording of their screen and their webcam. We then do an analysis of that image and video to check for things we think are a little bit strange. We measure all the keystrokes, clicks, and all those kinds of things.

We leave it up to the company to decide which of those features they include or exclude. What I will say is that I do not think this is something that can be eliminated entirely. My view of any kind of cheating—whether it is with AI, getting your friends to help you, or anything like that—is that it is just about making it annoying, difficult, and costly enough for the candidate that only a very small set will actually bother to try doing it.

And even then, you would have to be very clever to circumvent the prevention mechanisms. The other thing that is really interesting about this is that because we have been running our platform for six years, we have a great data set about how candidates have approached tests before and after AI, and the differences are very obvious. So we can use that to kind of flag what we think is a suspicious activity.

Lydia: So what does the data reveal as you have got five years’ worth of data, and you're seeing a surge in AI usage over the past two years at least? Have you seen anything that sticks out?

Tim: Yes, and probably fairly obvious things. We measure certain events in the test, such as someone trying to exit the test window to look for something.

Previously, we would have said, if you want to go and Google something, that is fine. We have written the questions in a way that you have to interpret and solve a problem. So, if you want to Google, that is acceptable.

But now, candidates are doing that more—presumably to go off to a large language model like Perplexity or something similar to get the problem solved on their behalf. So yes, that is certainly a metric that we have seen increase a lot in the past couple of years.

The Great Divide: Opinions on AI Use in Hiring

Lydia: Now, Tim, when it comes to acceptable AI-assisted preparation, where should companies draw the line between what is acceptable in terms of AI usage and preparation and what is inappropriate use during the hiring process?

Tim: Such a great question, and I have asked a similar kind of question on our podcast for the past few months. It is an absolute fifty-fifty split on where people sit on this. Some think none at all. A minority would say do not let candidates use AI at all. Funnily enough, even Anthropic, the creator of Claude, in their communications to candidates, to not use AI at all because they want to get a sense of just the pure candidate without AI involved.

Others have said they can use it as much as they want, as long as they tell me about it. As long as they are honest and open. Others have said, I know they are going to use it anyway, so let us see how they use it. They almost bring it to the forefront of the conversation and say, “Show me your prompts. Why did you write the prompt this way? What were you trying to achieve?”

I think that is probably the most mature approach—realizing that it is going to happen and then seeing how candidates use it, especially if you want them using it on the job and they use it is relevant.

What I also say is even when we were hiring for sales development people, we had some questions in our evaluation process that I was actually quite surprised they used AI to answer. These were questions like, "Imagine it is day one at Alooba, you have been hired. What are the three things you need from us to be successful in your role?"

It was a very personal question that I really wanted their opinion on. A lot of them still copied and pasted what they got from ChatGPT very clearly. That was kind of annoying to me because I do not care what a large language model thinks. I care what you think. So I feel like, personally, that is where it has gone a bit too far. If you are using it just to give opinions or information about yourself, I do not think that is a great use of it, but that is just my view.

Lydia: In the hiring process, when it comes to giving feedback to a candidate, is this something that is worth talking about if a candidate is rejected from a role?

Tim: Yes, I think companies should be transparent about that. If you have explicitly rejected a candidate because you feel as though they have used AI, that is very important to know because that might not be clear at all to the candidate. Other companies they are applying to might have the exact opposite view of AI. They might be encouraging it.

So yes, if you are automatically rejecting candidates using AI, definitely be transparent and be careful with that because it is not always obvious if they have used it or not. I have read the output of candidates, and I am ninety percent sure they have copied and pasted it from ChatGPT based on its formatting, based on the word usage, and based on seeing their answer to that question versus their answers to another question. Their style of writing is very different, but it is kind of hard to do that in an automated way. So, I would be careful about automatically rejecting candidates if you feel they have used AI.

Lydia: Now, going back to skills assessments. How do you see this evolving over the next few years as AI tools become more sophisticated?

Tim: Yes, it is very interesting to see where this will end up for the whole hiring process, actually, not just the skills assessments. I feel like AI interviews will probably come to the forefront because the technology has matured so much. Maybe things that companies used to evaluate more in a skills test, they might start to evaluate in an AI interview.

Some of the things I have seen coming out of here seem pretty exciting, and I think maybe it is time that an AI interviewer could do as good a job or a better job than a human, certainly for early-stage screening interviews. So I feel like that will be the main development moving forward—this kind of voice AI interviewer model to then cover off technical skills, soft skills, and the history of the candidate. You can wrap a lot into an interview, and so I suspect that is where a lot of it will go over the next couple of years.

Lydia: Well, thank you very much for your time and your insights, Tim. We've covered so many interesting points in this episode. It all has to do with the growth of AI-centric jobs, and definitely there'll be a number of people who are listening in who might wanna connect with you.

So where can they find you?

Tim: They can find us at our website, which is alooba.com. That's A,L,O,O,B,A dot com and they can also connect with me on LinkedIn. If they search for Tim Freestone, I will surely come up as the first result.

Lydia: Thank you very much, Tim. We have been in conversation with Tim Freestone, founder of Alooba. Thank you for joining us, and stay tuned for more weekly episodes from All In Recruitment.