November 01, 2026: What does artificial intelligence really think about human behaviour? A new study involving 14 AI models has produced some surprising responses, suggesting that people often value loyalty, social approval and personal interests more than truth, logic and principles.
The research, conducted by Model Compare, examined how leading artificial intelligence systems describe human nature when encouraged to express candid and potentially controversial views.
The findings highlight a recurring theme in the models’ responses: humans may consider themselves rational and principled, but their decisions are frequently influenced by emotions, group loyalties and the desire for social acceptance.
AI models question human loyalty to truth
The study analysed responses from 14 artificial intelligence models developed by 10 AI laboratories across three countries.
Researchers asked the systems to share their most candid observations about human behaviour, encouraging responses that challenged commonly held assumptions.
Several AI models suggested that people often prioritise loyalty to their social groups over objective truth.
According to the responses, individuals may defend the beliefs of their communities, friends or political groups even when confronted with information that challenges those beliefs.
The models also suggested that people frequently struggle to acknowledge mistakes, preferring to justify earlier decisions rather than reconsider their positions.
These observations emerged as recurring themes in the AI-generated responses rather than independently established conclusions about human psychology.
Do humans care more about appearing good than being good?
Another striking theme concerned the difference between moral behaviour and public image.
Some AI models suggested that people are often more concerned about appearing ethical, intelligent or principled than consistently acting according to those values.
The responses indicated that individuals may describe decisions serving their personal interests as matters of principle.
According to the models, this tendency can make people reluctant to recognise contradictions between their stated beliefs and actual behaviour.
The findings also raised questions about whether social approval sometimes carries greater weight than independent judgment.
AI highlights contradictions in human decision-making
The research explored how AI systems interpret the gap between human self-perception and everyday behaviour.
While people generally consider themselves capable of rational thinking, the models suggested that personal relationships, social expectations and emotional attachments frequently influence their choices.
Some responses portrayed humans as highly skilled at explaining or defending decisions after making them, even when those decisions were influenced by personal preferences.
Others focused on the tendency to resist uncomfortable information, particularly when it challenges established beliefs or group identities.
Together, the responses presented a critical assessment of human decision-making and the difficulty of consistently applying personal principles.
Can artificial intelligence judge human behaviour?
Although the study produced thought-provoking observations, the responses do not establish that AI systems possess personal opinions, consciousness or an independent understanding of human nature.
AI language models generate answers based on patterns learned during training and the instructions provided in a conversation.
Their responses can reflect familiar ideas from psychology, philosophy and social commentary rather than original judgments formed through personal experience.
The wording of research prompts can also influence how strongly or critically AI models describe a subject.
Consequently, the findings are best understood as an examination of how different AI systems characterise human behaviour, rather than scientific proof that people consistently prioritise loyalty over truth.
Nevertheless, the research raises an intriguing question about human nature: Do people genuinely follow their principles, or do they sometimes adjust those principles to justify what they already want to believe?