The Forum Network is a space for experts and thought leaders—from around the world and all parts of society— to discuss and develop solutions now and for the future. Register for free for updates, your own profile and to join the conversation!
The OECD's 2022 International Conference on AI in Work, Innovation, Productivity and Skills will take place 21–25 February—register now to bookmark sessions, meet the speakers and access special content!
Having a job matters. It puts food on the table, it pays the bills. It’s often a ticket out of poverty. Having a good job is even better: when people like their job or it matches their skills and interests, they tend to be happier and healthier.
For companies, finding workers to do the job is equally important. And finding good workers with the rights skills and attitudes is even better, because these workers tend to be more productive and less likely to leave.
But matching people to jobs is not as easy as it seems. There are millions of jobs and millions of workers out there. Finding out about opportunities, figuring out what the job entails, or what qualities a candidate truly possesses, is challenging.
In fact, this process can be so challenging that jobs remain vacant for longer than desirable, which raises the unemployment rate; or the wrong people are matched to the wrong jobs, which lowers productivity and increases turnover.
It’s no wonder that companies and policy makers get excited by the promise of new technologies, and especially artificial intelligence (AI), to speed up and improve the recruitment process.
Natural language processing that improves job descriptions, algorithms that target vacancies at the right candidates, vision and voice recognition software to carry out automated interviews and chat bots that take over much of the standard communication with candidates—these are just a few of the technologies that have emerged in recent years.
Hundreds, if not thousands, of companies and developers are selling these types of products, touting their potential benefits including lower costs, faster recruitment, reaching larger candidate pools, and more objective information to base decisions on, among others.
These tools might also improve the jobs of people involved in the recruitment process. Imagine if, instead of sifting through hundreds of CVs to screen candidates, workers could spend more time on interviewing and negotiating with applicants?
Read more: Rewiring the Firm: Algorithmic management and the future of work by Jeremias Adams-Prassl, Professor of Law, Faculty of Law, University of Oxford
A 2019 report by the HR Research Institute estimated that one in ten human resources professionals already makes high to very high use of AI in talent acquisition. Public and private employment services are also beginning to adopt AI tools, such as Pôle Emploi in France, or the VDAB in Flanders (Belgium).
Unfortunately, to date there is little robust evidence to back up the claims made by developers about the benefits of their AI tools, which are often little more than marketing strategies and advertising. Many of the assumptions on which their assertions are based remain untested.
However, some encouraging evidence is beginning to emerge from the use of AI in public employment services. In Korea, for example, job seekers had previously spent an average of 10 minutes searching for job-related information on various websites. Through the AI-powered The Work, however, candidates can obtain the same information within five seconds. In Flanders, training recommendations made by AI have been shown to lower the time job seekers spend in unemployment by 20%.
But there are still many unknowns, as well as potential risks. Everyone has heard of the famous Amazon AI recruitment tool that showed bias against women, because the model was trained using historical data on recruitments, which were mostly male. Significant doubts have also been raised around the accuracy of facial, voice and emotion recognition tools—especially when applied to certain sub-groups of the population.
In addition to bias, there are concerns that AI tools could dehumanise the recruitment process, and that they reduce a complex issue like human resources management to what can be measured.
There are also privacy concerns, as some of these tools scrape data on applicants from social media platforms and use them for unintended purposes.
Moreover, AI tools are a bit of a black box, making decisions and recommendations difficult to explain, thereby increasing the chances that something might go wrong. For the companies involved, the legal and reputational risks are very real.
Despite all this, human resource professionals appear optimistic about a future with AI. But it is also clear that the key to success lies in AI complementing, rather substituting, human decision-making.
Regulation may also play an important role in ensuring that the benefits of AI outweigh the risks. The EU recently put forward a proposal for new laws that identify the use of these technologies in employment as “high risk”. In order to be placed on the EU market, AI systems would have to comply with a set of mandatory requirements for trustworthiness and follow conformity assessment procedures.
In the United States, where policy makers tend to take a more reactive approach to regulation (as opposed to the EU’s proactive approach), some individual states have also taken steps. A law in Illinois requires applicants to be notified and asked for consent if their video interviews will be analysed by AI. The state of Maryland enacted a law that banned the use of facial recognition during applicants’ interviews, unless the interviewee signs a waiver. And, in November 2021, the New York City Council banned the sale of “automated employment decision tools” without annual bias audits.
Hopefully, with the right combination of regulation and best practices—as well as proper research and evaluation to better understand the risks and benefits—AI will contribute to more efficient matching in the labour market which, in turn, would result in lower unemployment and higher productivity.
See the full session line-up for the OECD's 2022 International Conference on AI in Work, Innovation, Productivity and Skills including a focus on AI for labour market matching—register now to join us!
|Digitalisation||Artificial Intelligence||Privacy & Cybersecurity|