The eliminative paradigm has run its course

Most ATS platforms operate on elimination logic. Keyword gates. Sequential screening. Multiple-hurdle filtering that discards a strong candidate because a single box fails to tick — regardless of compensating strengths elsewhere on the profile.

Operationally efficient. Informationally destructive.

77% of employers reported talent shortages in 2024 — the highest in nearly two decades. The same year, candidate satisfaction with recruitment continued its decline. Both sides of the labour market are unhappy with the mechanism that is supposed to match them.

Share

This is the architecture of adverse selection. When keyword gates determine access to human review, candidates learn to optimise for algorithmic legibility rather than authentic self-presentation. The signal degrades. The most employable candidates — those with options — withdraw first. The market repels the talent it most needs.

The problem is not the technology.

It is the embedded logic.

Personnel selection science has long established the alternative: a compensatory model in which strengths in one dimension offset gaps in another. Sackett and Lievens have argued this for two decades. The dominant ATS architecture has ignored them.

Tomorrow at IGCKM, with my co-authors at the University of Molise, we present

ARM extends the relational logic of CRM theory into the assessment process itself. The applicant becomes a stakeholder in a co-creative process, not an object of sequential filtration. Assessment becomes dialogue, not gate. Silver medallists become the next hire, the next referral, the next advocate — not records to be deleted.

The EU AI Act reclassifies recruitment AI as high-risk from August 2026. Penalties reach €35 million or 7% of global turnover. The systems most organisations still rely on were not designed for human oversight, transparency, or auditable decision rationale. ARM was designed for all three.

We have built the recruitment infrastructure of the previous century to industrial throughput specifications, then expressed surprise when it produces industrial throughput results. The eliminative paradigm has run its course. The relational paradigm begins.

Antonio Specchia is the author of Customer Relationship Management for Medium and Small Enterprises (Routledge, 2022) and the forthcoming Applicant Relationship Management: Human-AI Collaboration in the Agentic Economy (Routledge, 2026). He maintains an active research programme with the University of Molise and is based in London.