Why do people suddenly sound more robotic when an AI is judging them?
Because they assume that is what the machine wants. It is a predictable reflex, and it is working against most candidates who do it.
When you learn that an algorithm will parse your resume, score your one-way video, or analyze your voice, something shifts in how you present yourself. You start choosing words that sound logical and measured. You trim the stories that reveal warmth or instinct. You assume a machine rewards data and penalizes feeling, so you give it data.
Researchers have a name for this. Across twelve studies with 13,342 participants, Goergen, de Bellis and Klesse documented what they call an "AI assessment effect," published in the Proceedings of the National Academy of Sciences in June 2025. When people believed AI was assessing them, they emphasized analytical characteristics and downplayed their intuitive and emotional ones. The driver was a lay belief that AI prioritizes analytical traits. In other words, the behavior change is not based on how the systems actually score people; it is based on what candidates imagine the systems value.
Does sounding more analytical actually help you get selected?
Usually not. The same research warns it can change who gets hired, and not in your favor.
The authors of the PNAS study, writing in Harvard Business Review in July 2025, cautioned that this self-presentation shift could fundamentally alter who gets selected and undermine the validity of the assessment process. The reason is straightforward. When you strip out empathy, creativity, and intuition to seem more machine-friendly, you are erasing the very human qualities that distinguish outstanding employees from merely adequate ones. You are not optimizing your candidacy; you are flattening it.
There is a second layer to the mistake. A peer-reviewed study in Scientific Reports, published in July 2025, found that candidates wrongly assume AI is worse than humans at judging interpersonal skills, so they downplay those skills during AI selection. The encouraging finding is that this misperception is correctable. When people were simply informed that modern AI techniques can assess interpersonal traits, the mistaken belief faded. So if you have been hiding your relational strengths because you assumed the algorithm cannot see them, the premise itself may be wrong.
What happens when you try to game the system with keywords and scripts?
It tends to backfire on two fronts: the system can flag it, and the human reviewing your recording can feel it.
Recruiting and career experts have pushed back hard against over-correcting for the machine. In reporting compiled by Murray Resources from Wall Street Journal coverage in February 2026, voices including Keith Wolf of Murray Resources, J.T. O'Donnell of Work It Daily, and Ben Sesser of BrightHire warned that candidates often fall into sounding scripted or robotic and try to keyword-stuff their answers. That is a mistake, they note, because systems can detect overly robotic answers, and a human ultimately reviews the recording. You are rarely performing for the algorithm alone. You are performing for the person who will watch your video after the algorithm shortlists it.
Think about what a stilted, keyword-dense answer communicates to that reviewer. It signals someone who is managing an impression rather than answering a question. It reads as anxious, not analytical. The candidate who tells a clear, specific story in their own cadence almost always lands better than the one reciting a checklist of competencies.
How should you actually present yourself when AI is part of the process?
Present yourself the way you would to a thoughtful human interviewer, and trust that your full range is an asset rather than a liability.
Start by naming the reflex so you can resist it. If you catch yourself editing out a story because it sounds too emotional or too instinctive, that is precisely the moment to leave it in. The research suggests your instinct to suppress is the trap, not the safeguard. Speak in complete thoughts at a natural pace. Use concrete examples with real outcomes, and let your reasoning show, including the moments where judgment or empathy shaped a decision. Those are not weaknesses to hide from a scoring model; they are signals of the qualities employers say they want most.
When you describe how you solved a problem, include why it mattered to the people involved, not only the metrics. When you explain a decision, you can say you weighed the data and also trusted your read of the situation. That blend of analytical and human is what actual job performance looks like, and it is what a competent reviewer is listening for.
This matters more now than it used to. In a June 2026 Harvard Business Review article, Shraddha Sunil and Mudit Saraf argued that generative AI is rapidly eroding the reliability of traditional hiring signals like polished resumes and structured interview answers. As polish becomes cheap and easy to manufacture, authentic specificity becomes the differentiator. A genuine voice, a real example, a flash of how you actually think: these are harder to fake and easier to trust.
The takeaway is not that analysis is bad. It is that performing a narrow, machine-pleasing version of yourself sacrifices the qualities that make you worth hiring. The candidates who do best are not the ones who guess what the algorithm wants. They are the ones who show up complete.