Services
People
News and Events
Other
Blogs

Can I use Artificial Intelligence in recruitment?

View profile for Employment Team
  • Posted
  • Author

Artificial Intelligence (AI) is becoming increasingly common in employee management and recruitment. There is no universal definition of AI, but for the purposes of this article, AI refers to computers that use algorithms to analyse large amounts of data and make decisions much more efficiently than a human could. Our Employment Law team discuss the potential benefits and drawbacks of using AI in recruitment and things you should keep in mind if utilising AI in your workplace.

Uses and benefits of artificial intelligence in recruitment

Employers can use AI throughout the recruitment process to perform tasks such as:

  • Sifting through CVs;
  • Searching candidates’ social media for certain terms;
  • Analysing tone of voice and facial movements in interviews;
  • Online assessments and tests.

Using AI for recruitment can greatly increase efficiency, reducing both the time and cost spent on recruitment. Computers can scan hundreds of CVs and compile a short list of qualified candidates much faster than a human manager. Using algorithms can also reduce the influence unconscious bias can have on the process. For example, if you use a hybrid system where AI is used for part of the recruitment process you can ensure that gender bias (where the recruiter prefers one gender over another), confirmation bias (where the recruiter makes certain assumptions about a candidate or their CV based in their own experiences) and name bias (where the recruiter has a preference for names like their own) are avoided.  If you are considering implementing AI in your business, it is important that you first identify your reasons for doing so and what you expect the AI to achieve.

Drawbacks of using artificial intelligence in recruitment

One potential drawback of using AI is that it can dehumanise the recruitment process which may put some applicants off. It may also make it more difficult for managers to develop relationships with new employees as they may not have had a chance to get to know them during the recruitment stage.

Also, it is argued that AI may discriminate against applicants by replicating human biases. There have been several examples of well-intentioned AI resulting in biased decision-making. One example is Amazon’s algorithm for screening candidates’ CVs. The algorithm was taught to screen CVs using past recruitment data for Amazon going back ten years. By analysing this data the algorithm “learned” that women applicants were less preferable and therefore penalised CVs which contained words or phrases that indicated the applicant was a woman.

Another study found that advertisements for vacancies for careers in maths, science and engineering were more likely to be shown to men on platforms such as Google and Facebook. The reason for this related to the fact that young women are a more valuable demographic for advertisers and ads targeting them are therefore more expensive. To minimise costs, the algorithm tended to target men over women.

The above examples illustrate how biases can make their way into AI in unexpected ways. One way to safeguard against this is to ensure that humans oversee the process and have the final say on any decisions. However, some managers may become too reliant on AI, and feel compelled to accept its decision without performing any critical analysis of their own. This may also lead to an overall decrease in transparency and accountability for recruitment decisions.

GDPR considerations when using artificial intelligence in recruitment

Article 22 of the UK GDPR restricts employers’ ability to automate decision making by giving people the right, “not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her.” The exceptions to this are where the decision is:

  • necessary for entering into or performing a contract between the data subject and data controller;
  • required or authorised by UK law; or
  • based on the data subject’s consent.

You may be able to use AI to make recruitment decisions if one of the above exceptions applies and you have in place additional safeguards to protect the data subject’s rights and legitimate interests, including the right to have a human intervene in the decision and the right to contest the decision.

You may also be able use AI for decision making where the decision is not solely based on automated processing, such as where a human actively reviews the decision and makes the final call on whether or not to apply it.

Recommendations for employers regarding artificial intelligence in recruitment

You should be transparent about any use of AI or automated decision making in your recruitment procedures and inform applicants of their right to object to its use or contest the decision. Managers who use AI should be trained on how it works so they can effectively oversee the process and evaluate any decisions. These managers should then have the final say on whether to apply the decision made by the AI.

Artificial Intelligence and algorithms are incredibly useful tools, but whether they are “good” or “bad” all depends on how they are used. As these tools continue to expand and develop we may see further developments in the law regarding their use. If you have questions about the current legal position regarding the use of AI in recruitment or other areas of employment law, contact our Employment Law Team on on 023 8071 7717 or email employment@warnergoodman.co.uk.

To receive regular Employment Law updates from the team regarding recent tribunal cases and legislation updates, you can subscribe to our weekly Employment Law Newsletter by completing our subscription form or emailing us at events@warnergoodman.co.uk

ENDS

This is for information purposes only and is no substitute for, and should not be interpreted as, legal advice.  All content was correct at the time of publishing and we cannot be held responsible for any changes that may invalidate this article.