Most discussions about AI in recruitment focus on the employer side: screening, scoring, ranking, and automated decision-making. This white paper looks at the other side of the labour market, the candidate side. And why AI should not only be judged by how it screens people, but also, when well deployed, by how it helps them access opportunities.
This white paper takes as its reference point the February 2026 Belgian study Recrutement et sélection à l’ère de l’IA, commissioned by the Institut pour l’égalité des femmes et des hommes and conducted by researchers from Liège University (LENTIC, HEC-Liège) and Hasselt University (School voor Sociale Wetenschappen).
It examines how AICCA, an AI-assisted career-guidance platform used by jobseekers and employment support professionals, compares with the risks, concerns, and recommendations identified in that study.
The distinction is important. The systems examined in the Belgian study are used on the employer side of the labour market, to attract, screen, assess, rank, or select candidates. AICCA operates on the candidate side. It does not evaluate people for recruiters or employers. It helps jobseekers and the professionals who support them prepare stronger, clearer, and more usable job-search materials.
That changes the nature of the AI question.
Many of the most serious concerns around AI in recruitment arise when systems are used to make or support decisions about individuals with limited transparency and weak human oversight. AICCA is structured differently. It does not score, rank, filter, or reject candidates. It generates draft materials that remain under full user control, whether used in guided sessions with a support professional or directly by the jobseeker.
The paper also argues that candidate-side AI should not be viewed only in terms of risk avoidance. When properly designed and responsibly deployed, it can help reduce a disadvantage that already exists before any recruiter-side system comes into play. For many jobseekers, especially those with limited literacy, weaker command of the local language, or less access to high-quality support, poor application materials can create a real penalty at the very first stage of access to employment.
By helping users produce clearer, better structured, and more linguistically correct materials, while preserving human review, user agency, and transparency, human-centred AI can contribute to fairer access to opportunity.
This white paper sets out that case, reviews AICCA against the findings of the Belgian study, and highlights several ways in which its design aligns with, and in some respects goes beyond, the safeguards and principles now being called for in the debate on AI in recruitment.
Key takeaways from the paper:
- AICCA does not make decisions about individuals. It does not score, rank, or reject candidates.
- Human oversight is built into the model. In guided use, the support professional is in the loop by design. In autonomous use, the jobseeker remains fully in control.
- Transparency is not optional. Users are explicitly informed that results are AI-generated and should be reviewed and edited before use.
- Prompt governance matters. AICCA relies on centrally designed and controlled prompts, rather than ad hoc personal prompting.
- Candidate-side AI can help counteract exclusion. By improving the quality of application materials, it can reduce disadvantages that disproportionately affect more vulnerable jobseekers.