It’s been a year since Generative AI burst onto the scene and took over the entire TA conversation for 2023. Some of the predictions made back then about its immediate impact on the world of work were out there, even for something that clearly has so much disruptive potential.
So, one year on, where have we got to? What are the genuine use cases, what is still hype, and how will AI actually transform talent acquisition over the next few years?
In the interview, we discuss:
• How AI has evolved in the last six months
• Slow gains but clearer potential
• Generative AI versus Conversational AI
• What AI can and can’t do in the context of Talent Acquisition
• Where is adoption the quickest?
• TA process re-engineering
• What will things look like in two years and five years?
• Emotional work versus repeatable tasks
• Implications for recruiters and their careers
• AI and Job Seekers
• Advice TA leaders on AI strategy
• What progress can we expect in 2024?
Matt Alder: Support for this podcast is provided by Paradox, the conversational AI company that’s transformed the hiring process for global TA teams like General Motors, Nestle and McDonald’s to get recruiting work done faster. What if recruiting and hiring was as simple as this? Yes, this, the conversation I’m having with you right now. With Paradox, it is. They leverage conversational AI to seamlessly automate time consuming recruiting tasks like applicant screening, interview scheduling and onboarding via chat conversations right on the candidate’s phone or laptop, so recruiters can spend more time with people, not software. Paradox has helped global employers save countless hours and millions in recruiting costs simply by making hiring simple and conversational. Visit paradox.ai to learn how Paradox can work for you. That’s Paradox dot AI.
[Recruiting Future theme]
Matt Alder: Hi there. This is Matt Alder. Welcome to Episode 576 of the Recruiting Future podcast. It’s been a year since Generative AI burst onto the scene and took over the entire TA conversation for 2023. Some of the predictions made back then about its immediate impact on the world of work were a little bit out there, even for something that clearly has so much disruptive potential.
So, one year on, where have we got to? What are the genuine use cases, what is still hype, and how will AI actually transform talent acquisition over the next few years?
My guests this week, Adam Godson, President and Chief Product Officer at Paradox and Yael Florenthal, VP of Product at Paradox, are the perfect people to answer these questions.
Matt Alder: Hi, Adam. Hi, Yael. Welcome to the podcast. An absolute pleasure to have both of you on the show. Please could you introduce yourselves and tell us what you do?
Yael Florenthal: Yeah, with pleasure. So hi, everyone. My name is Yael Florenthal. I’m a VP of Product at Paradox. I’m located in our Tel Aviv office. Producting stuff, exciting stuff. A lot of focus recently about Generative AI, which we’ll talk about soon.
Adam Godson: Yeah, great to be with you again, Matt. I’m Adam Godson. I’m President and Chief Product Officer at Paradox. My goal is to push boundaries in talent acquisition technology and then make all that happen.
Matt Alder: Fantastic stuff. Now you and me, Adam, we last spoke in March about Generative AI, because there’s not been that much else to talk about this year other than [chuckles] Generative AI when it comes to talent acquisition. My first question is, obviously, the pace of change is very, very fast here. What’s changed since March? How has it evolved, and how has the way that you’re using generative AI in practice evolved as well?
Adam Godson: Yeah, it’s interesting. I think the overall biggest story that I would say is that, there have been slow gains but not breakthroughs in how it’s changed since March, where I think companies have started to convert a little bit of the potential energy, which everyone sees into kinetic energy of getting some things done. But I don’t think there’s a use case I can point to that says, “There’s been a giant breakthrough in talent acquisition where we no longer have to do this thing, or there’s this giant push and breakthrough in this other way.”
I think we’re starting to see some pushes forward, some interesting copilot models, some interesting conversational AI, and we’ll talk about that as we go forward. I think we’re seeing some interesting tool sets. But I think a lot of it is still very early days where it’s left to the swivel chair integrator, the person to use three or four different tools to get stuff done, not really built into our systems yet. But I think some of the potential energy itself even is getting clearer about what will the use cases be, how will we use this, what are some of the challenges? So it’s all getting clearer. But I think it’s in some ways, maybe been a little slower than people might have expected it to be.
Matt Alder: Yeah, I suppose after that kind of initial, I can’t even think, almost hysteria when this launched. I suppose before we get into any more detail, it’s probably worth just getting some definitions down here in terms of, what’s the difference between generative AI? Paradox talks a lot about conversational AI. There are other types of AI as well. Talk us through all of them and how they work.
Yael Florenthal: So basically, both generative AI and conversational AI, they both stem from the same umbrella of artificial intelligence, but they serve different purposes. So generative AI is all about creating new content. It can create text, it can create images, even code. For example, you can ask for it to generate a job advertisement for you, and it can do it from scratch, just simply understanding language, and the World Wide Web and create that job advertisement for you.
Conversational AI, however, is designed for human interaction. So, a very common example is your typical AI chatbot on a website where it’s programmed to understand human queries and then respond in the way that actually meet that criteria and is based on an existing database. So, while they’re both under artificial intelligence, the purposes are different for each one.
Matt Alder: And what about things like data? Because people often talk about AI, and data and all that kind of stuff as well. Where does that fit in?
Yael Florenthal: With artificial intelligence, AI, you are asking? With generative AI, it is when you ask for it to generate new content for you, it is based on an existing set of data. It might be some like ChatGPT, which is based on a very wide range of data, and thus can create very exceptional and very creative answers. You can also host the data yourself and run the generative AI on your own data. So basically, it all comes down to what are you looking to achieve, and what is the task that you want to complete. And based on that, you will carve out the data you want to inject the generative AI.
Matt Alder: I suppose that leads on to my next question, because we tend to talk about generative AI as a big term, and this is what’s going to revolutionize talent acquisition. What parts of it are the best fit for TA, and what bits aren’t actually that useful or that useful at the moment?
Yael Florenthal: So one example I can think about that’s probably what Adam is thinking about as well is it really excels at automating and streamlining very repetitive tasks, such as creating job descriptions. Or, I don’t know, sorting through resumes, or labeling candidates or conversations or data. So this doesn’t only save a lot of time, but it also ensures some consistency. It can potentially reduce unconscious bias, at least in the early stage of recruiting. So that’s one task, for sure, it’s really good at.
Matt Alder: Is there anything that isn’t useful that people are maybe sort of focusing on the wrong thing when they’re–? Because I know lots of people are experimenting with it. I had a podcast interview yesterday with their head of TA and their recruitment marketing team, and they were doing all kinds of experimentation with different types of generative AI. Is there people going off in the wrong direction? Are there things they shouldn’t be focusing on?
Yael Florenthal: I think it’s really exciting, this technology. There has been a lot of hype and maybe still is. It can solve a lot of stuff, and it could really help us with those routine tasks. I know it can’t really fully grasp complexities of human interactions or human emotions or interpersonal dynamics. There’s probably a lot of startups there that are trying to solve it, but it’s not really there yet.
So if we’ll take an interview, for example, and you have a candidate that maybe shares an example about a conflict they had in their previous job, the AI can maybe understand the sequence of events and it can outline it, but it can’t really grasp the emotional aspect of it. It can’t really learn what we can learn when we see a person shares about a conflict.
Matt Alder: That makes a lot of sense. So how is AI helping your clients right now? What kind of recruiting challenges are you solving with the AI that you’re using? Is there a particular type of company that you are helping the most? How’s it looking right now?
Adam Godson: Yeah, I think we talked some months ago. I think we focused on making a generative conversational AI experience, and we wanted to focus on that part to how do we make Olivia, who’s our conversational assistant, have a great interactive two-way conversation. And so I think the conversation quality is the part where we’ve raised the bar significantly this year, and working with our teams to understand what are the guardrails. And the bar for quality in this industry has to be very high. Candidate ask the question, “What are the benefits for this role and you get them wrong?” [chuckles] We’ve got to be right. And so how to reduce errors and hallucinations, how to be sure that we get the right information based on that user.
Because the potential here is about one to one personalization to be able to personalize every conversation to the individual having it and the company, they’re having it with. But there’s a lot of challenge to making that happen. And so for us, it is improvement in all the metrics around conversational quality. So we get more people to say thank you. We get better completion rates, we get more questions asked per conversation, we get better explicit ratings of those conversations, we get less confusion in conversations. They all overall just feel warmer and better to objectively have those questions get answered in a recruiting process and then have that work get done.
Matt Alder: In terms of type of companies, are there particular areas of industry or types of hiring where the adoption is quicker than others?
Adam Godson: I think we’ve mostly seen companies that care a lot about their employer brand. And so sometimes that’s consumer brand companies folks that have external product brands and folks that are willing to push some boundaries, and are oftentimes using AI in other parts of their business or have groups that are using AI. They send us lots of long questionnaires about [chuckles] AI safety and all those things as you can imagine. But it’s people that are willing to push boundaries and maybe have a little experience with that.
And there are many other companies that are willing to let other companies go first, find where the boundaries are in the world, and they’ll go second. And that’s fine too. It takes all types.
Matt Alder: So difficult questions now. [chuckles] I’m going to split this question into two because I’m really interested to differentiate. So I know it’s obviously, it’s impossible to predict what’s going to happen. We’ve talked about, there’s been no big breakthrough since March. By the time this goes live, in a week or so’s time, [chuckles] the world-
Adam Godson: We’ll see.
Matt Alder: -the world may have changed. But with that caveat, how do you think things are going to develop? First of all, in the medium term, what do you think is going to happen over the next two to three years with all of this?
Adam Godson: I think what ultimately has to happen is process reengineering of all of our talent acquisition processes. And so today, what we’re doing is we are using generative AI in the current processes, so figuring out how that can help certain points in those processes. There’s a copilot model, there’s a toolbox model, there’s conversational agents. There’s lots of ways to do those things. But it’s all based on the frameworks we’ve had for 50 years of how we do talent acquisition. And so I think what really has to change in the next step is a reengineering of how people think about the recruitment process and how the entire process will change.
I think that starts to separate the types of work into transactional work and emotional work. I use the word emotional intentionally there that it really is the type of work that requires human depth. I mentioned it earlier about interviewing someone and understanding their story about a conflict. I also sometimes talk about the act of a recruiter of convincing someone to join. That is emotional work that technology cannot do. It cannot convince you to leave your job, move across the country, interrupt your career for something, and join you at their cause, because there’s no cause. They’re not there. They’re not real. And so that emotional work will have to be done by a person. And so that process at reengineering to separate that emotional work that humans need to do versus the transactional work that can be automated.
Matt Alder: I want to come back and talk about that in terms of the implications for recruiters, and careers, and all that stuff. Before we do though, super impossible question. Where do you think we’re going five years down the track? I know that there was a lot of speculation recently that OpenAI might have invented artificial general intelligence or something incredibly scary. Where do you think we go in five years’ time with this?
Adam Godson: Yeah. I think the recruiting process in five years looks completely different than it does today. I think the way that people are assessed for roles, the way that they apply for roles. Another good example is actually the job seekers in many ways are ahead of companies at this. So today, we’ve got a dynamic where job seekers are using AI to write their resume or their CV. I’ve seen estimates as high as 60% to 70% of CVs are now being written with AI tools.
And on the other side, we’ve got companies using AI tools to read them. [chuckles] There eventually is a breakdown that starts to happen around, how do we even assess who this person is at scale and new methods may be needed to do that. And so I think in five years, the talent acquisition process looks entirely unlike what it does today and how we think about assessing people for jobs, how they go about raising their hand to jobs.
And ultimately, I think there should be celebration. I don’t think there’s anyone that looks at today’s process and says, “Wow, that’s an amazing process. This is how I would design it.” [Matt laughs] That’s where I think we’ve got a responsibility and an opportunity as an industry. We actually do get to think about how we start to design this thing as we would have it and start to make that happen. That’s an incredibly exciting time.
Matt Alder: I think you’re right. I think if you transported someone in a time machine from 30 years ago and asked them to apply for a job, they would recognize so much of the process, resume, cover letter, face to face interview, all that kind of stuff, which shows how far we haven’t come. I’m really interested as to when that point of lots of AIs talking to each other and why are we doing this anymore, we need to reinvent things if things comes. I think that’s really interesting.
I suppose, combining all that together, lots of people listening will be thinking about what their future career might look like in talent acquisition if as things are going to change like this. What does this mean for recruiters and what do you think their role is going to be? I suppose more importantly, what kind of skills are going to be needed in two-years’ time, three-years’ time, five-years’ time, however long it is?
Adam Godson: Yeah. I think there will be a bifurcation of the skill set and what people want to do. I think today, we conflate a lot of what people consider recruiting work as not that value added. There’s a lot of clicking work. There’s not a lot of recruiting going on. It is a lot of moving people in systems, and administrative work, and clicking through the ATS, where a real recruiting work is about convincing people to join. It’s emotional. It is about how do I convince you to come to my company, how do I be sure that you’re the right fit? I think we can automate a lot of the transactional conversations now. That’s one of the things that will change with AI is some of the things like a job intake conversation or a screening conversation, those things could get automated.
And so I think from a skill perspective, there’s going to be a choice where people decide they have high emotional intelligence. They love talking to people. They want to convince people to join every conversation of the day. They will absolutely love that. Those are the recruiters today that really hate the system stuff. That stuff all just gets in the way of talking to people that’s what they love. And then there’ll be people that focus more on the systems aspect. And so I think you’ll see some divide there and you’ll see a lot less transactional work.
Matt Alder: You mentioned earlier about people going first and people wanting to be second to see what happens with implementing these kind of things. What would your advice be to heads of talent acquisition who are looking at how they integrate AI into their strategy moving forward? On the one hand, what should they be cautious about? What should they be looking out for? But then on the other hand, what do they need to do to make sure they stay ahead of the curve?
Yael Florenthal: I can take that. So I think I know that now more than ever, AI Tools has become much more user friendly and accessible than ever. So first of all, I advise anyone who hasn’t done so to go to claude.ai or to ChatGPT and just start experimenting and start writing prompts. You really don’t need much to start writing prompts and asking it to perform tasks for you. There’s a lot of magic to this combination of words, generative AI, as if again, it could solve all of your problems. It is a very smart technology and it’s really good. If, for example, you want to customize your email, that’s one thing you can do it easily today within a matter of like 10 seconds. But it’s up to you. So if you’re going to ask it to write a romantic poem for your 15 years anniversary and decide what you should get your partner, that’s very risky. So it’s up to you.
You need to lay your guardrails in advance and decide how far you’re going to take it, and create constant feedback loops, and make sure it’s working for you, and you’re ahead of the game. So pick your poison or pick your problems to be solved wisely is what I would suggest.
Matt Alder: So final question. We talked about the medium-term, we talked about the long-term. We’ve talked about a point which I think is very clear that the talent acquisition is going to have to change. How does this journey start? What happens in 2024? What progress do you think we’re likely to make in the next 12 months? If were having this conversation again this time next year, how would we review the year?
Adam Godson: Yeah, it’s great. I hope to have this conversation next year. Let’s get it on the book, Matt.
Adam Godson: I think 2024 is going to continue to be a year of experimentation of people understanding a certain part of an existing process, how can I get some gains, typically in a toolbox a method? How do I solve a single problem in a simple way for part of this? And that’s actually a great way to start.
For example, one of the things we see our clients start with is using AI for interview scheduling, rescheduling, even the complex stuff. There’s oftentimes a human person or group of people, and frankly no one loves it. [chuckles] But finding some discrete tasks to really automate and use AI to do that and do that well and get those games and build confidence. We typically see people go from average time the schedule of four days to 30 minutes, just some really big change in those where they can then point to a win, go to get more money [chuckles] and say, “Let’s continue to automate and use AI in that.” So I think we’ll continue to see lots of wins in discrete single problems.
My hope is that we’ll start to rethink some processes more deeply and think about how we can change the way that talent acquisition operates. I think it’ll take some certain industries and some certain problem sets to find some of that creative thinking in that. But I fully expect 2024 again to be a more with less kind of a year, where recruiters are feeling strapped and they’re looking for tools like AI that can help and can help them have more of those meaningful conversations that they’re looking to have.
Matt Alder: Adam, Yael, thank you very much for talking to me.
Yael Florenthal: Thank you.
Adam Godson: Thanks, Matt. Great talk as always.
Matt Alder: My thanks to Adam and Yael. If you’re a fan of the Recruiting Future podcast, then you will absolutely love our monthly newsletter, Recruiting Future Feast. Not only does it give you the inside track on what’s coming up on the show, you can also find everything from book recommendations to insightful episodes from the archives, and get first access to new content that will help you understand where our industry is heading. For a limited time, subscribe to the Recruiting Future Feast newsletter and get instant access to the video recording of the recent remixed webinar on AI and talent acquisition, featuring some of the smartest thinkers in the industry. Just go to mattalder.me/webinar to sign up. That’s Matt Alder dotme slash webinar.
You can subscribe to this podcast on Apple Podcasts, on Spotify, or via your podcasting app of choice. You can find and search all the past episodes at recruitingfuture.com. And don’t forget to sign up for the newsletter, Recruiting Future Feast. Thanks very much for listening. I’ll be back next time and I hope you’ll join me.