Nvidia's chief software architect Jonathan Ross said job seekers could gain an advantage by using the same artificial intelligence model that recruiters use, citing new research that suggests AI hiring systems may favor resumes generated by their own models.
Nvidia's chief software architect Jonathan Ross said job seekers could gain an advantage by using the same artificial intelligence model that recruiters use, citing new research that suggests AI hiring systems may favor resumes generated by their own models.
"Someone did a study and proved that resumes generated by one language model (LLM) are rated higher by that model than resumes created by another artificial intelligence," Ross said during a speech at the Sohn 2026 investment conference, Business Insider writes .
"Now recruiters use LLM to decide who to invite for an interview. So you just need to find out what language model a particular recruiter is using," he added.
Ross noted that candidates may need to have several different resumes tailored to different AIs to maximize their chances of passing automated selection systems.
"So you should create one resume using Claude or Opus 4.7 and the other through ChatGPT. That way you essentially have the highest probability of getting selected," he said.
It seems that Ross was referring to a recent scientific paper titled “Selfish Biases of AI in Algorithmic Hiring,” published in late 2025.
Researchers tested over 2,200 resumes in 24 different professional fields and found that candidates who used the same AI model as the evaluator were 23–60% more likely to be shortlisted than applicants who submitted human-written resumes with similar qualifications.
These remarks come at a time when AI-based recruiting tools are rapidly spreading among corporate recruiting departments.
According to a 2025 Resume.org study of nearly 1,400 U.S. employees who are familiar with their companies’ hiring processes, 57% of organizations have already integrated AI into their hiring process. Of those employers, 79% use AI to screen resumes, while 74% reported that automated systems are capable of rejecting candidates without human input.
