He has been warning about it for years. A.I. can offer excellent solutions for many recruitment tasks, but you should never place too much trust in it, says automation specialist Martyn Redstone. This is particularly true when you ask LLMs to rank candidates. He refers, for example, to ‘rank roulette‘: today’s number 2 can easily be number 5 or even lower tomorrow, meaning the outcome is different every time. And perhaps worse still: A.I. recruiters rarely agree with one another. In only 14% of cases do their recommendations more or less match.
‘LLM screeners reinforce the inconsistency that already plagues selection, rather than remedying it’,
‘LLM screeners reinforce the inconsistency that already plagues recruitment and selection, rather than remedying it’, Redstone therefore concludes. ‘Until that volatility is tamed, all claims about the “fairness” such systems are said to bring are nothing more than marketing smoke.’ A.I. is really a kind of ‘overconfident intern’, he says: ‘quick and smooth, but shockingly inconsistent’ in its assessments. Good candidates also sometimes simply ‘disappear’, and the A.I. systems also remarkably often use the same reasoning points in their assessments.
Read this article in full
Become a Werf& member for unlimited access to all articles, research and exclusive insights into the labour market.
No obligations Cancel anytime Secure payment