Using artificial Intelligence (AI) for recruiting can enhance your candidate selection process, but beware of ‘black box’ algorithms that can lead to recruitment decisions that you can’t defend. Richard Justenhoven at cut-e explains the two types of AI system and how it can be used effectively in assessment.
Artificial Intelligence (AI) is often perceived as a futuristic technology, but the truth is that it is already part of our lives. AI creates your music and movie recommendations, suggests possible online purchases and protects you from fraudulent activity on your credit card.
AI can now improve your organisation’s employee selection process too, by enabling you to assess many more applicants. It can analyse and interpret a huge volume of candidate data quickly and cost effectively. But caution needs to be applied here, because the basis of your selection decisions must be legally defensible.
Two types of AI system
There are two types of AI system: those that use neural networks and custom systems.
Neural network AI systems utilise ‘deep learning’, so they ‘learn’ as they analyse large volumes of data – and they adapt their behaviour accordingly.
The fact that the algorithms in the system identify patterns and trends, and respond to them, sounds promising but what this actually means in practice is that the system will reject certain candidates – but you won’t know why. The system won’t give you a reason for its decision, so it’s almost impossible to understand why some candidates are accepted and others are not.
A ‘black box’ problem
This is called a ‘black box’ problem. Recruiters, particularly those in regulated industries such as financial services, should always know the basis for any selection decision. If you cannot justify exactly why a candidate has been rejected from your application process, you’re vulnerable to a legal challenge from that individual. Every organisation has to be able to legally defend its selection decisions.
Another problem with plug-and-play systems is that they all do the same thing in the same way. In other words, they won’t differentiate your employer brand. If your competitors are using the same systems, you’ll all reject the same candidates – and you’ll all be chasing the same talent.
Custom AI systems are different. These systems continuously learn by ‘observing’ the best practice of human raters. When a number of human raters assess a candidate’s responses, looking for specific behaviours and competencies, a custom AI system can learn to understand exactly what criteria they use to rate people. To avoid any possible issues of bias from an individual rater, the system will only match those aspects that are universally agreed on by all of the raters. Through this ‘teach by example’ approach, the system will then begin to rate other candidates in exactly the same way.
The advantage here is that your selection decisions will be legally defensible, because you can explain exactly how and why they were made. However, the downside of custom AI systems is that it can take up to six months to ‘train’ them to assess candidates in exactly the same way that your human raters would judge them. You have to ‘pre-feed’ the system with relevant information, so that it can learn to replicate the best practice of your raters. Managing this lead-time will be a major challenge for organisations.
Benefits of a project team
If you want to utilise AI for recruiting, the best way forward is to form a project team now to evaluate how a custom AI system can enhance your video interviewing and other recruitment processes.
The sooner you do this, the sooner you’ll benefit from AI’s ability to quickly analyse vast quantities of candidate data and deliver useful information that will support your selection decisions. If you decide not to do this, you’ll always be six months behind those pioneering companies that are investing now in AI technology.
Download ‘Artificial Intelligence in assessment’ – a free white paper by cut-e