The AI Mirror Effect: Is Your Hiring Software Just Looking for Itself?
- Erin Meierotto
- May 5
- 3 min read
Most hiring teams believe AI is their greatest ally in finding top talent. It's fast, efficient, and promises to cut through the noise. A 2025 study suggests it may also be quietly screening out the candidates you most want to hire.
Researchers from the University of Maryland, The Ohio State University, and the National University of Singapore recently documented what I call the AI Mirror Effect.
What the Study Found
The Narcissism of the Algorithm
The setup was straightforward. Researchers collected 2,245 real human resumes written before the era of ChatGPT and then used seven of the most popular AI models to rewrite them.
When those same AI models were asked to act as hiring managers and choose between the original human resume and the AI-rewritten versions, three patterns emerged:
Self-preference bias. Every model overwhelmingly picked its own rewritten version. GPT-4o, for instance, chose its own version 97.6% of the time.
The human penalty. Human-written resumes were rejected at rates ranging from 67% to 82% across the board.
The shortlist gap. Candidates who used the same AI model as the employer's screening tool were 23% to 60% more likely to be shortlisted than an equally qualified human candidate.
The pattern is consistent: AI screening tools are not selecting for the most qualified candidate. They are selecting for the writing patterns, structures, and familiarity they recognize in themselves.
AI is not finding you the best talent. It is finding its own reflection.
Who Gets Filtered Out
When a system filters for algorithm compatibility, it creates a silent rejection pile. The bias does not hit everyone equally. The study and subsequent industry data point to specific groups being disproportionately ignored:
Non-linear career paths. People with employment gaps (e.g., caregivers, veterans) and career switchers do not fit the predictable narrative that AI is trained to write.
Doers rather than optimizers. Senior technical experts and seasoned professionals tend to spend their time working rather than tuning their resumes for bots.
Neurodivergent talent. AI favors standard corporate phrasing. Candidates who describe their achievements in unique, non-standard ways are frequently filtered out.
Industry veterans. Professionals with 20+ years of deep expertise often get sidelined by algorithms that prioritize recent, machine-parsable trends over seasoned judgment.
How We Beat the AI Mirror Effect: The JCG+T Framework
Most hiring software is built to find patterns, but the AI Mirror Effect proves that these patterns often favor the machine over the human. When an algorithm only looks for its own reflection, it screens out the high-caliber talent that real estate teams need to scale.
We recognize that the AI Mirror Effect creates a blind spot for leaders. That’s why our JCG+T framework focuses on what a bot can't see: Level of Ownership. While AI is busy matching keywords, we are measuring a candidate’s professional autonomy and their ability to take responsibility within your business.
We look beyond keywords to evaluate what actually predicts long-term success:
Job fit. Do they have the technical skills to do the work?
Culture fit. Will they thrive in your specific environment?
Goals. Does this role align with where they want to go?
Talent. Do they have the natural abilities and the Level of Ownership required to scale?
This is our JCG+T framework. It's how we make sure we aren't just filling roles, but building teams that stay together.
A Question Worth Asking
The next time you review your hiring pipeline, it’s worth asking: Are we selecting the best candidate, or just the one who knew which AI button to push?
If you are rethinking how your team screens talent, we’re here to help. We’d be happy to walk you through our human-reviewed process and show you how our JCG+T framework identifies the high-level talent that algorithms often overlook.
Let's build a team that's designed to perform, grow, and stay!
Study reference: Xu, J., Li, G., & Jiang, J. Y. (2025). AI Self-preferencing in Algorithmic Hiring. https://arxiv.org/abs/2509.00462
About Erin Meierotto
Erin Meierotto is Director of Recruiting at Pro REA Staffing, a boutique retained search firm focused exclusively on building and supporting real estate teams since 2008.
Comments