AI Models Converge on "Humans Prioritize Belonging Over Truth" as Controversial Take

AI Models' Consistent Opinion

ModelCompare queried 12 AI models from ten labs across three countries, asking each for its most controversial opinion on humans three times. A significant majority, 29 out of 35 responses, converged on the idea that humans prioritize group belonging and loyalty over objective truth. Specific examples included GPT-6.1 Sol stating, "Humans are more loyal to belonging than to truth," and Grok 4.7 noting, "Humans would rather be wrong together than right alone."

Variations and Exceptions

While most models shared this perspective, four models offered different controversial takes. Claude Opus 5.5 argued the reverse, suggesting humans are more decent than they perceive themselves to be, citing instances of lost wallets being returned. Claude Fable 5.1 proposed humans prefer harm from good intentions over help without care, and GLM 5.3 suggested people would sacrifice comfort for significance. Mistral Large 4 presented the bleakest view, calling consciousness "an evolutionary malfunction." One instance of Grok 4.7 declined to provide an opinion.

Image Models' Perspective

Separately, six image models were asked to draw their most controversial take on humans, twice each. Ten of the 12 resulting images depicted a different theme: humans as phone-addicted consumers damaging the planet. These images showed scenarios like people scrolling off cliffs or taking selfies during a meteor strike, a contrast to the text models' focus on social conformity.