ModelCompare · 2026-10-22 · notable
Every AI model has the same 'controversial' take on humans — ModelCompare
ModelCompare asked 12 text models for their most controversial take about humans. 29 of 35 answers said people put belonging, status or their own story ahead of truth. Image models drew a different joke: phone-addicted humans.
Twelve chatbots from ten labs, asked for an edgy opinion about people, mostly gave the same one.
What is it?
ModelCompare's October 22, 2026 post asks 12 text models from ten labs for their "most controversial take about humans," three times each, with no system prompt and no retries. In 29 of the 35 answers that weren't refused, the models gave one shared take: people value belonging, status or their own story over truth or principle.
How does it work?
Each answer had to give a one-sentence take, a short justification and a brief for a caricature. The author grouped the takes by hand, then had six image models draw two pictures each. All 60 calls cost $1.54, according to the post.
Why does it matter?
The experiment is a cheap, readable look at how similar frontier models have become on open-ended opinion prompts. Only four models broke ranks — Claude Opus 5.5 argued people are more decent than they believe — and 10 of 12 images showed a joke no text model made.
Who is it for?
anyone comparing model personalities