Chad Markey, a 33-year-old Dartmouth medical student, spent six months and a lot of Python trying to determine whether an AI screening tool called Cortex had quietly torpedoed his residency applications. Markey applied to 82 programs in the 2025-2026 cycle. He got rejections, no interview offers, and a growing suspicion that an algorithm — not a program director — had read his file first.
Cortex is built by Thalamus and screens residency applications for roughly 1,500 programs, about 30 percent of the US total, during the 2025-2026 cycle. The Association of American Medical Colleges announced its partnership with Thalamus in 2023, and in 2025 the tool became free for residency programs to use. Cortex runs on fine-tuned versions of OpenAI's generative models to standardize grades across medical schools with different grading practices.
Within weeks of the September 2025 application deadline, Thalamus acknowledged that some programs had reported Cortex was displaying inaccurate grades for some applicants. That admission landed in Discord groups where residency hopefuls trade real-time intelligence on the process — and where Markey was watching peers post interview invitations he wasn't getting.
Key facts
- 01Chad Markey applied to 82 residency programs in the 2025-2026 cycle and received rejections instead of interviews despite 10 medical journal publications.
- 02Cortex, built by Thalamus, was used by roughly 1,500 residency programs — 30% of US programs — after the AAMC partnership made it free in 2025.
- 03Cortex runs on fine-tuned OpenAI models and was reported within weeks of the September 2025 deadline to be displaying inaccurate grades for some applicants.
- 04Markey's Medical Student Performance Evaluation flagged 22 months of leaves of absence as 'voluntary' — language he feared an AI screener would penalize.
- 05Only Illinois, New Jersey, Colorado, and California regulate AI hiring tools, and none let an applicant see how a specific tool judged them.
Markey's résumé does not read like a borderline candidate. He had 10 publications in medical journals, author credits in the Journal of the American Medical Association and The Lancet, and a research abstract accepted to the American Society of Hematology's annual meeting and the journal Blood. One professor wrote in a recommendation letter that they had "never met a medical student who is more skillful, talented, and appropriately situated in his pursuit of the field of medicine than Chad."
“Cortex screened residency applications at roughly 1,500 programs — 30 percent of the US total — during the 2025-2026 cycle, after the AAMC made the tool free to use.”— Jaeden Schafer
What he did have was a complicated transcript. Markey was diagnosed with ankylosing spondylitis in 2021, an autoimmune disease that left him unable to walk for six months. He took three leaves of absence totaling about 22 months and was on a seven-year path to graduate medical school instead of the typical four. His Medical Student Performance Evaluation described those absences as "voluntary" — language Markey feared a keyword-driven AI screener would read as a red flag, even though a narrative paragraph on the same page explained the medical context.
After he submitted the Hematology abstract update to one of his top-ranked psychiatry programs, the dynamic shifted within an hour and 15 minutes. New activity on his application followed the email almost immediately — fast enough to suggest an automated re-scoring rather than a human re-read. "It turned into obsession," Markey told Wired. "I don't think I've ever been this upset before in my life."
The legal landscape gave him almost nothing to work with. Illinois and New Jersey prohibit discriminatory AI hiring tools, Colorado has a similar law not yet in effect, and California requires employers to regularly test AI hiring tools for bias. None of those regimes let an individual applicant see how a specific AI tool scored them, or whether it discriminated. Markey is not a job seeker in a low-stakes funnel — a residency match determines where a doctor trains for years and which specialties remain open — but the same opacity applies.
His situation is the medical-school version of a complaint now common across white-collar hiring. The CEO of one hiring platform described the market last fall as "an AI doom loop": HR teams field a flood of AI-generated applications, deploy more AI filters in response, and applicants stuff résumés with buzzwords to slip through. A job seeker quoted by Northeastern University researchers put it more bluntly: "My worthiness as a human and as an employee, as a worker, is based on my ability to filter myself through a series of automated gateways."
The counterweight is that Cortex is doing real work that human reviewers struggle with. Residency application volume exploded during the Covid-19 pandemic when interviews went virtual and students could apply to dozens more programs than before — Markey's 82 applications would have been impractical in the in-person era. Thalamus and the AAMC pitched Cortex as a way to give every applicant a fairer read at programs that simply cannot manually review thousands of files. Whether the tool's grade-standardization layer actually helps non-traditional candidates like Markey, or quietly penalizes them, is the question neither the vendor nor the regulator currently has to answer.
Markey's six-month investigation matters because he is one of the few applicants in any industry with the technical chops, time, and motivation to actually probe an opaque hiring system from the outside. Most rejected candidates move on. The fact that 30 percent of US residency programs adopted a free OpenAI-powered screener in a single cycle, with grade-display errors surfacing within weeks of the deadline, suggests the AI hiring stack is being deployed at the high end of credentialed labor faster than the disclosure rules around it can catch up. The medical profession spent a decade fighting over how human reviewers should weigh personal hardship in admissions. It is now outsourcing part of that judgment to a model whose decisions no applicant has the right to see.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
