If the machine chooses

AIR: AI in Recruitment Discussion tool

When you choose to implement artificial intelligence systems within recruitment and selection processes, you have to do it in a very intentional way. And with a keen eye on the underlying values. The AI in Recruitment (AIR) Discussion Tool can help you do that.

It consists of 20 questions within the themes of fairness, data use, autonomy, bias, and validation. By thinking and discussing these questions as a project group when implementing AI, you can avoid key risks.

AI in recruitment infographic

This infographic shows where artificial intelligence is being used in the recruitment and selection process. This software already makes regular independent decisions. For example, about who gets to see a vacancy, whose resume fits the vacancy well enough, or which candidates' soft-skills best fit the model of the ideal candidate. Increasingly, it is the machine that chooses

The use of AI in recruitment

In the recruitment and selection process, organizations initially try to get as many suitable candidates to apply (sourcing) and then choose the most suitable candidate from those (selection).

For both of these phases of recruitment, there are artificial intelligence systems on the market that can help organizations with the process. These technologies have an impact on who is or is not selected and thus must be used with extreme care.

Benefits

Potentially, this type of technology has a number of benefits. For example, it could help reduce bias within the process. Irrelevant person characteristics can be automatically disregarded, and you can measure in what ways the system is biased a lot easier than with a human recruiter. Technology could also help find new groups of candidates who were not previously in the picture.

In addition, the benefits of using artificial intelligence have mostly to do with efficiency. Parts of the process can be automated, and the process can be made more uniform.

Risks

However, there are also significant risks associated with the use of artificial intelligence within recruitment and selection processes. Because much of the technology assumes the current (successful) workforce, there is the chance that you are actually maintaining the (not very diverse) status quo.

It can never be ruled out that very specific forms of bias regarding certain groups remain in the system, and these types of systems are poor at dealing anyway with individuals who deviate in some way from the norm. That bias that remains - even after careful implementation - is then, however, immediately systematic and scales with the deployment of the technology.

Furthermore, it remains difficult to validate whether the artificial intelligence you deploy is working properly.

Finally, these types of systems require a lot of data. This can pose problems in terms of privacy and required data minimization.

It is already often the machine that chooses
You often hear that we need not worry about deploying artificial intelligence within recruitment and selection. After all, for now, it is still humans who make the final decision. This may still be true (for now) for the hiring process, but it has long since ceased to be the case for the candidates who are rejected. There, it is often already the machine that chooses, without any human intervention.