How has the rise of ChatGPT and off-the-shelf LLMs disrupted AI recruiting tools?

After ChatGPT emerged, hundreds of tools and startups began offering AI-based recruiting and matching systems, making previously expensive capabilities accessible to more companies.

What legal and bias challenges have emerged around AI scoring tools in recruiting?

New York State, Illinois, and several other jurisdictions passed laws requiring companies to prove that their AI scoring or inference technology is not discriminatory.

What was Eightfold's pioneering contribution to AI-powered recruiting?

Eightfold was a pioneering company that scraped profiles of individuals from across the internet, anonymized them, and used AI to study skills and technical capabilities.

Do AI recruiting tools favor resumes written by the same AI model?

August 29, 2026 The Josh Bersin Company

Yes. Research by computer scientists examining hundreds of resumes found that AI recruiting tools using ChatGPT are 50 to 75% more likely to accept resumes also written with ChatGPT compared to those written by hand or with competing models like Claude or Gemini. This bias occurs because the tools recognize the specific word formations and embeddings produced by their own language model.

The finding exposes a troubling problem at the heart of modern hiring: as Josh Bersin explains in The Josh Bersin Company podcast, we now have "AI creating resumes and producing and sending them out and AI consuming them — slop talking to slop." The recruiting industry has entered a feedback loop where both candidates and hiring platforms rely on the same technology.

How AI model preference distorts the screening process

The bias isn't deliberate discrimination—it's an artifact of how language models work. When ChatGPT generates a resume, it produces specific linguistic patterns and numerical embeddings unique to its training. A recruiting tool built on the same model naturally recognizes and resonates with these patterns more readily than text generated by a different system or written manually.

This creates an invisible advantage for candidates who use a particular AI tool to write their resume. A candidate might be equally or even more qualified, but if they wrote their resume by hand or used a different model, the screening process discussed in this episode will rate them lower simply because of the mismatch between how they expressed themselves and how the AI evaluation tool was trained.

A systemic problem in recruiting scale

The stakes are enormous. Recruiting is a massive industry—20 to 30% of Americans change jobs every year, and hiring a single highly-skilled person costs between $5,000 and $10,000 or more. For senior roles, companies may pay nearly one-third to one-half of an executive's first-year salary to recruiters who find and vet candidates.

When AI recruiting tools systematically favor resumes from their own ecosystem, qualified candidates are filtered out at scale, and hiring decisions become shaped by technology preference rather than merit. This dynamic is explored in depth in The Josh Bersin Company, where the implications for fairness and hiring outcomes are made clear.

Language model embeddings: Numerical representations that capture the semantic meaning of words and phrases. Each language model (ChatGPT, Claude, Gemini) produces unique embeddings based on its training data. Recruiting tools recognize patterns from their own embeddings more easily than those from other models, creating an inherent bias toward content generated by the same system.