As we have seen in this chapter, selection decisions rely on data gathered from applicants using a variety of methods. For decades, humans have used these data to consider its match to job descriptions and person specifications in an attempt to predict who will perform well in a job. But as we have also seen in this chapter, the pace of technological change is rapid and Artificial Intelligence (AI) promises (or threatens depending on your point of view) to revolutionise selection processes.

While many AI processes are some years away from being mainstream, they are nevertheless reasonably straightforward adaptations of the electronic Applicant Tracking Systems that many organisations already use. The scale of investment in the field is huge: there are currently hundreds of HR-tools using AI and machine learning (the capacity of systems to improve from experience without being programmed to do so) and large and small tech firms alike are pumping millions into improving these and developing more. There are many current examples. First, software for the automated analysis of video interviews that tracks facial expressions to indicate suitability in relation to body language and tone of voice. Second, psychometric matching programmes that analyse application form completion and eliminate applicants using too many backspaces/delete as this might indicate lack of decisiveness. Third, systems that tailor psychometric tests to needs of job, delivering 70% reduction in recruitment time and improved performance of new employees. Others include predictive analytics about likely performance of applicants and chatbots to conduct initial interviews. The potential for automation is seemingly endless and in many cases is demonstrating positive results. For example, Unilever has adopted AI in its (volume) graduate recruitment using gamified psychometric testing, AI analysed video interview and algorithm-driven selection process. 80% of new appointments are judged to be performing well in role and it has the most diverse group of new recruits ever. AI may, however, be less effective for specific, niche roles.

In many ways, AI is to be welcomed, given the widespread recognition of the subjective nature of many current selection processes. AI algorithms can be completely objective and offer the opportunity to eliminate bias, creating diversity in selection as Unilever has found. Care is nevertheless needed as initial programming is done by humans, who should take care not to perpetuate their own (often unconscious) biases in selection algorithms. Care is also needed as the machines learn: if their analysis of current top performers identifies these as young, white men (possibly because of the shortcomings of current selection processes) then these characteristics will be built into developing algorithms and perpetuate bias. Human oversight will be needed, although it may in time be possible to train systems to recognise and ‘correct’ for potential biases.

Ultimately there are two possible scenarios: humans use and are informed by technology and AI or AI takes over and makes people largely redundant. The first one is probably more likely, given that the applicant experience is important and human involvement is fundamental to this. Intuition is also still considered to be important, although this can be controversial. Nevertheless, AI will change HR roles, possibly for the better as the more routinised aspects will be removed, but numbers of jobs will almost certainly reduce. HR needs to rethink its role in the selection process and how it will continue to add value.

Questions

1. What are the benefits and draw backs in using AI in selection?

2. How accessible is AI to firms that are not large and well-resourced?

3. How will AI change the shape of HR activity?