Welcome to episode 3 in our ‘The Future at Work’ podcast! Here we listen to Prof. Chris Forde, Co-Director of the Digital Futures at Work Research Centre to discuss key debates in the use of AI in the recruitment process. For more information, read our blog here!

Ellen: Hello everyone, and welcome to the ‘Future of Work’ podcast series my name is Ellen Wang, Faculty International Manager here at Leeds University Business School and your host for this podcast series. In the last two episodes, me and my co-host Marc have discussed the broad prospects and trends on adaptation of AI in recruitment. We've also invited industry practitioners to share her view on this topic. In today's episode will focus on some of the academic debates around the use of AI in recruitment and selection. Looking at key areas and debates, for example, the use of AI in recruitment selection, is it inevitable and what challenges does it present to organisations? And also, what are the implications for workers, etc. Now joining me today, we have someone who's an expert in this area of research, Chris Ford, who is a professor of employment studies in Leeds University Business School, and deputy director of the ESRC Digital Futures at Work Research Centre. This centre, involving over 70 researchers at the University of Leeds, Sussex and others in examining the future of work and the role of technology in the workplace. Chris has research interests around the platform and gig work as well as in the areas of HRM, recruitment and selection. So, welcome, Chris. Thank you very much for joining me today for this fascinating topic. Chris: Thanks very much, Ellen.

Ellen: So Chris, let me start by asking you how widespread is the use of AI in recruitment and selection in your research? And why do you think this is needed or even being used, please?

Chris: Well, the use of technology in recruitment and selection, things like the tracking of applicants or technology to gamify interview tasks, that's quite widespread. There was a recent survey by the Chartered Institute of Personnel and Development, which found about half to 2/3 of organisations were using such technologies in their recruitment and selection processes. But with artificial intelligence, so when we're talking about the use of visual perception by computers, decision-making by computers, perhaps less so, it's an aspiration for many firms. I think rather than a reality, so that same report by the Chartered Institute for Personal Development, found actually only about one in seven firms were using, artificial intelligence and automation in any area of the human resource processes, and that's much smaller than the use of artificial intelligence in areas such as production, manufacturing and information technology.

Ellen: And why do you think this is? I mean the use of AI is often portrayed as inevitable and or desirable for firms. Isn't this the case?

Chris: I think you’re right, it’s really good question. As with many functions that involve administrative tasks, human resources has activities have got very ripe for picking in the use of artificial intelligence. So, recruitment is often put forward as a primary candidate for artificial intelligence, it can be used to help source and rate applicants in a fair and equitable way, but also artificial intelligence might be used in areas of learning and development, field and employee enquiries. So, that all sounds really positive and you'd expect it to be used quite a lot in the area of HR, but in the area of recruitment selection there is a lot of challenges to do in that it's not as straight forward as it might seem, so first of all, it's costly. It's a lot to invest in these artificial intelligence systems. One estimate suggested that even for a small enterprise, if they were trying to introduce the artificial intelligence system in to their recruitment selection, it might cost them anywhere between 50,000 – 100,000 thousand pounds start-up costs just to implement that technology. Then, in addition to that, you've got trying to train applicants and train managers, most importantly in order for it to be effective. So that’s the first thing. But secondly, I think perhaps most importantly the adoption of these technologies is not, it's not just that technically neutral exercise or decision, and it’s a social process. Something which is driven by social processes, and power relations in the workplace and maybe adopted for a whole host of reasons. So, a paper by Deborah Hambercroft and Phil Taylor, just published this year, makes this point very clearly. They know that artificial intelligence is it really expanding in the area of HR. And, some organisations are using it simply to improve efficiency, but many organisations and managers within those organisations have quite other motives for introducing artificial intelligence. They want to try and increase control over the workforce, perhaps are trying to reproduce existing power relations and dynamics in the workplace, and these are definitely not neutral decision. Colleagues at Leeds University Business School, David Spencer, Mark Stewart, Simon Joyce and Xanthe Whitaker, have looked at this in a recent report for the European Parliament, and they know also that the decision to adopt artificial intelligence in areas like recruitment and selection is shaped by a whole host of factors. The regulatory environment that is in place in the UK, for example, sectoral context, you know some sectors and types of organisations just have more of a tradition of invested in the sort technologies. We look at health and social care, for example, there's a lot of interest at the moment in the use of intelligent technologies to assist in areas of healthcare delivery. In that sort of environment, perhaps there's more current incentives and pushes for managers and organisations to invest in artificial intelligence in other areas. So, I think there are challenges to introducing it. It is not simply something that can be done in a short term way and without consideration of the broader environments and context in which these technologies are being introduced. And just finally at a practical level, not everything in recruitment selection can be automated or should be automated. Think about during COVID research done during the COVID19 pandemic has revealed that many job applicants do value that human interaction they want to sit next to someone face to face and speak to them. This is an organisation that might be going to work in, and they want to understand and combat human interaction with people and rather than just sort of robotic interaction.

Ellen: Right. Thank you, Chris. That's really interesting. In particular, I have picked up some figures, I am quite sensitive to the figures that you said. Some organisations that are required to invest somewhere between 50 to 100K. So my next question really is, I mean for those organisations who invested in such technology, what is their return on investments? I suppose you know, the question is, what do you think the benefits are? Because in the previous two episodes we discussed from students perspective, and practitioner perspective, so I’m really interested to hear what does academic research suggest from your perspective please.

Ellen: So are we saying first impressions count anymore?

Ellen: Absolutely. That's really fascinating. And, just when you say that I was thinking about things like do, do they, I mean the artificial intelligence, or the algorithm, pick up overtime the tone of voice and the confidence in the voice. And you're not all that kind of stuff. That maybe system learning that or machine learning that could be, it could be done, as a way of not reducing bias and fairness, but actually maybe increasing bias and fairness, right?

Chris: Yeah, it's possible. It is definitely possible. And I think the more, these technologies are getting more intelligent, I think so the artificial intelligence systems in recruitment selection can recognise particular voices, tones and can try and make judgements and decisions based around that, so they are definitely improving and it does depend upon what they are programmed to do though. If your organisation is, and it really just poses a question about what the organisation is looking for when it's recruiting and selecting. If it is recruiting and selecting on objective criteria, perhaps these can be built into artificial intelligence systems to come up with objective decision but if the organisation is seeking more, subjective attributes and behaviours. It may be a real challenge to build these into artificial intelligence systems, and indeed some would say, there's a there's a lined to be drawn some point about the extent to which technologies or human recruiters should rely upon particular attributes or behaviours. So, I think it's about thinking about this as a HR human resource is person, I would try and think about this in the in the round as artificial intelligence being just one part of a well-designed recruitment and selection process. It perhaps has something that can help systems to overcome obvious biases and limitations. I think that's definitely a positive thing, but alongside other human aspects of recruitment and selection processes as well. So I think it's a balance between it's between those two things I think.

Ellen: Sure, definitely. So given these gaps in the technology, why can't the technology behind these artificial intelligence systems being improved, to overcome these obvious in limitations do you think?

Chris: I think one of the key challenges related to the human dimension of recruitment and selection. Ultimately, the reality is that the hirer is still a human, a manager, a person in an organisation or people within organisations. So, despite the promise of fairer recruitment through artificial intelligence, will managers really trust this technology to undertake recruitment selection? Will they relinquish control of recruitment and selection and give it over to these artificial intelligence technologies? Now interestingly, this was a key question posed in a recent piece of research for this Digital Futures at Work Research Centre that I'm involved in, by Will Hunter and Jackie O'Reilly and they looked at Walmart, one of the largest retail organisations in the world, and their use of artificial intelligence in recruitment and selection during the pandemic. They had a number of reasons for trying to produce artificial intelligence into their retail stores during the pandemic. They wanted to reduce face to face contact with applicants, which is deliberate decision. They had already been trying this before that pandemic, and they notice prior to visit they they had real issues with retention rates. So shop floor retail workers had relatively high turnover. So they were trying to reduce this. And one of the key things that they were looking for this technology to do was to identify applicants that would stay with Walmart for at least three months. So, the algorithm they used was designed to rank candidates based on their likelihood of staying in post for at least three months. Very data driven. And you know, it was implemented during COVID-19. I think in terms of its rankings, Walmart was very happy with the data that was being provided and the candidates that would be placed at the top of the list by the algorithms. But managers implicitly distrusted these artificial intelligence systems. So, interviews at the researchers did found that actually managers were starting to try and override the findings of the technology and many of them had reservations about whether this algorithm was actually able to identify good hires. So, what we saw was some managers moving candidates back down the list, even though the artificial intelligence algorithm had rated them excellent in a pre-employment assessment. Or conversely they found candidates further down the list to rated poor by the technology, who they actually thought would make good associates, so they were using their hunches, their intuitions, and making comments like, well, ‘I just don't trust the computer’, you know, I know what it's like to hear someone’s voice. The algorithm can't call people up and hear what they've got to say. I've got that human intuition about what this person will be like. And so they ended up bypassing the recommendations in some cases, so think overall you've got this potential misalignment between the algorithms recommendations, and the use of his own judgments of quality based upon what they think someone's expected performance is going to be.

Ellen: Right. Yeah, that's very interesting. I mean, a picked up one keyword while you were sharing this with us is the control, so you know I think this is quite concerning is who's in control during the recruitment and selection process. Is it the system that we've created humans have created or is it the decision maker? At the end of the day, it’s a human looking for a job, working for human. The system is just a tool to I guess, to streamline the process and to make it fair. I don't know if, you know, what I'm saying is a correct? So I guess, given the research that you pointed out, to say that recruiters sometimes make poor and unfair decisions. What do you what do you think about this? You know this point in terms of control?

Chris: Yeah, not I think you. I think you're exactly right. I think it is. There is tension between recruiters wanting to improve the decision-making and using technology to do that, but also wanting to retain some control over this. I think it's about a balance again and this is coming back to this importance of trust. Walmart in this research were very interested in the findings of this research, and they tried to take steps to increase the transparency and the confidence in the system. So, they've tried to adapt the artificial intelligence to so that the user interface highlights the qualities and explains why candidates have been put nearer to the top of the list, so that a recruiter can have more confidence, perhaps in why a particular candidate is at the top rather than just seeing that someone's appeared at the top and having to, and having to trust the system is right. I think some further work will definitely needed to convince some users, managers, recruiters of the value of the system, and I think this does come down to then a culture of and the climate in which these technologies are being used. I think managers, recruiters, employees, job applicants will become more confident and trusting of technology, and if it's been seen in in other contexts and it's not simply being used to replace humans in these sort of activities. But it's being used to compliment the human decisions that are being taken. I think that's one of the big sort of conclusions here it's it's often portrayed I think in some of the popular press about this is robots are going to be taking over everyone's jobs, and we're going to be seeing artificial intelligence everywhere. And it is there is a lot of potential for artificial intelligence. I think in work and employment, but there are also limits and there's also ways in which humans and artificial intelligence can complement each other in complex areas of human resources like recruitment and selection.

Ellen: Sounds like a collaboration between a human and the AI system.

Chris: And yeah, I think you're right.

Ellen: Okay. Great. Well, thank you so much, Chris. I think this is really insightful for listeners. You know, especially that you shared some of the case studies with real organisations. So, as we're coming to the end of this episode, I just wanted to ask the big question, what are the big takeaways here from your perspective, please?

Chris: I think the first thing is that the adoption of these technologies, particularly artificial intelligence in recruitment, so actually is not straight forward. So it's not a quick fix solution and it's definitely dependent upon good programming, good algorithms and human intuition and intelligence about what the organisation is looking for in particular applicants. I also think it's really important to try and connect this to other aspects of the human resource cycle, if you like. If the aim is to, for example, in the recruiter the Walmart example. If the aim was to recruit employees who will be retained in and still be there in three months’ time, what's the rationale? What's the other short human resource policies and practises which are needed to try and ensure that that objective and that target is met. So, rather than just thinking about artificial intelligence in recruitment and selection, what else is needed to try and ensure that people remain with an organisation, does it come down to training activities, onboarding activities, retention policies? So I think it's this to me flags up really important issues about human resource is and the value of human resources within organisations, so I think there is real potential with artificial intelligence recruitment selection, but it also doesn't need to connect to other areas of human resource activities as well. The other things I think are really important takeaways here are the thing I mentioned at the start of the recording about the adoption of technologies and artificial intelligence being socially driven, that is shaped by social processes, organisational cultures, and beliefs and mind sets of managers, as well as broader institutional regulatory environments. So, it definitely is the case, I think that more firms are going to be pushed and encouraged to use artificial intelligence for recruitment and selection, but it's not straight forward and ultimately there's still a human element in there, still based upon human decisions, as we saw in the example. Walmart and managers will still have a role to play. I think finally thinking about from the job applicants perspective or the workers perspective, I think finding out as much as you can about potential organisations use of artificial intelligence is really useful exercise. If you are looking for a job, for example finding out how the organisation you're thinking of applying for is its using artificial intelligence in recruitment and selection, is really important for making an application. But it may also tell you something about the organisation that you're applying for as well, and you can understand and find out more about how it communicates with applicants, whether it's investing in artificial intelligence in other areas of its operations, what it's using these technologies for and for what purposes? I think these are definitely useful insights into organisations and how they operate in the 21st century.

Ellen: Absolutely. Well, there you have it our audience research, communication, these are the repeated messages that hopefully coming through from our episode, I think you obviously from academic perspective, researchers really, really important. Thank you once again Chris for sharing insights with us. But research not only just from academic research perspective, but also for our students, graduates. You know, if you are looking for a job right now, it is really important to research an organisation as a whole, their culture, you know their structure, their governance, all of which contributes to what we are talking about, that in terms of adaptation of AI in recruitment and selection. As Chris pointed out, there are other elements, it’s not just a machine, but all these algorithms that considers social processes and organisational cultures as part of it. But at the end of the day, I guess it's not a placement, as he said, complementary to human decision, right.

Chris: Yeah, I agree.

Ellen: Well, I guess that's all we have time for today. Times flies by when you’re having fun. So I guess, you know, for the next episode we will tackle on a completely different theme and subjects, but until now we've covered just three episode on adaptation AI in recruitment selections from student’s experience, practical perspective, industry and also some of the academic debates. And I hope that we shared enough information and insights for you to really understand and get hold of what the current situation is, and in order to prepare you for the future. So as always, if you're interested in finding out more about this topic, please subscribe to our podcast series. Also, welcome to get in touch with us or Chris himself by sending an email or booking an online appointment. Our contact details available in episode below until next time take care.