Estuaire Service
Contact us

HomeBlogArtificial intelligence

AI in recruitment: risks of bias and good practices for employers

Screening CVs, writing job adverts, preparing interviews: AI can help recruiters, provided they manage the risks of discrimination and keep the decision human.

Artificial intelligence3 min readFor companies, job seekers

A sales job advert published in Bafoussam can receive, within a few days, more applications than a human resources manager can seriously read. It is tempting to hand CV screening, or even candidate ranking, over to artificial intelligence. Some tools already offer this.

AI can indeed save time in a recruitment process. But it can also reproduce, and sometimes amplify, injustices that are hard to detect. For an employer, the challenge is twofold: recruit efficiently and treat every candidate fairly.

Where AI really helps the recruiter

  • Writing a clear job advert: duties, required skills, conditions, without jargon or exclusionary wording.
  • Building an assessment grid: objective criteria linked to the job, weighting, related questions.
  • Preparing interviews: behavioural questions and role-play scenarios suited to the job.
  • Summarising information: a synthesis of the notes taken by several panel members.
  • Writing replies to candidates: respectful rejection messages, interview invitations, requests for documents.

In these uses, AI plays a supporting role. It makes no decision about a person, and the recruiter checks every document produced.

Where biases come from

A bias is a systematic tendency to favour or disadvantage certain profiles for reasons unrelated to the real ability to do the job. With AI, it can appear in several ways:

  • the model learned from data that reflects past inequalities, for example jobs long held mainly by men;
  • the criteria given to the tool are poorly chosen, such as requiring a specific school or neighbourhood of residence;
  • the tool penalises unusual career paths: career changes, career breaks, experience in the informal sector;
  • it misreads CVs written in English, in imperfect French or with a simple layout;
  • it gives too much weight to keywords at the expense of real skills.

The danger is that these biases remain invisible. A table of scores gives an impression of objectivity, while the criteria that produced it may be questionable.

Take an example. A distribution company asks a tool to rank candidates for a storekeeper position, giving priority to those with “similar experience”. The tool then favours CVs that mention the word storekeeper and rules out a candidate who managed the stock of the family shop for five years without holding that title. Yet her skills matched the job perfectly.

Good practices for fair recruitment

  1. Define the criteria before reading the applications, directly linked to the duties of the job.
  2. Remove information unrelated to the job when you use a tool: age, photo, family situation, origin, religion.
  3. Never let AI eliminate a candidate on its own: a person reviews the rejected applications, at least by sampling.
  4. Test the tool: submit equivalent fictitious profiles that differ only by first name or gender, and compare the results.
  5. Inform candidates when automated tools are involved in processing their application.
  6. Protect personal data: a CV contains sensitive information that must not be copied into just any online service.

What this changes for candidates

For candidates, knowing that some recruiters use screening tools is a reason to make your CV easy to read: clear headings, explicitly named skills, dated experience, a readable file. There is no point, however, in filling your CV with keywords unrelated to your background: an attentive recruiter will notice, and it harms your credibility.

On a platform like Estuaire Emploi, a complete and honest profile remains the best way to be noticed by the right employers, whether or not they use automated tools.

In conclusion

Recruiting shapes the future of the company and of a person. AI can make the process faster and better organised, but only if the employer stays vigilant about the criteria, checks the results and owns its decisions. Fair recruitment is also more effective recruitment: it helps you spot talent that overly rigid filters would have ruled out.

Need a hand with recruitment or HR?

Estuaire Service handles recruitment, payroll, digital and training for companies in Cameroon.

Talk to our team