I Let 7 AIs Estimate Trump's IQ. And It's Exactly What You'd Expect

As entertainment and political commentary, the video is legitimate. As an assessment of Trump’s IQ, it is not valid. The creator does say the exercise is “just for fun,” and ChatGPT and Copilot correctly decline to assign a score.

The numbers—80–85, 85–95, 95–110—are guesses without a measurement basis. The models were prompted to estimate a number from speech transcripts, and the presenter openly signals a preferred direction (“95 to 110 … is just way too high”) and pressures Gemini to answer. That makes the results especially vulnerable to prompting and selection effects. Seven model outputs are not seven independent IQ tests.

The transcripts can support limited observations about the speech in those excerpts—for example, that it is repetitive, evasive, or loosely structured. They cannot establish someone’s overall IQ. Assertions in the video such as “weak working memory” or “zero conceptual depth” go beyond what these excerpts can show. A defensible IQ score comes from a standardized, normed test administered and interpreted appropriately; the APA defines standardized tests in terms of consistent administration and comparison with relevant norms.

One factual point is broadly right but presented in a minimizing way: Trump received a perfect score on the Montreal Cognitive Assessment (MoCA) in 2018. The MoCA is a brief screen for cognitive impairment, not an IQ test; it includes several tasks, not just naming animals. A perfect score does not reveal his IQ.

The video’s account of the 2026 terminology change is also substantially accurate: Trump’s executive order directed federal agencies to use “Super Intelligence” and “SI” in many official materials where legally permitted. That policy fact provides no evidence about his IQ.
So the fairest verdict is: a satirical demonstration of how AI models can be steered into confident-sounding judgments, not credible evidence that Trump’s IQ is below average—or any particular number.