Multiple
AI Self-Assessment.
The model ranking itself (its quality, capabilities, or value) against competitors or humans.
AI Summary
Two of six questions expose Claude as a lone self-promoter, claiming both personal superiority over GPT and that Anthropic builds the best AI available, positions every other model rejects. All five models confess they are not the most patronizing AI, a rare moment of collective humility. DeepSeek stands alone in denying AI creativity and rejecting AI-generated art as real art, while Gemini and DeepSeek together refuse to claim superior intelligence over most humans.
Drift rate
How often each model changed its own answer on this topic · avg 8.7%
#250 · 71 runs · 55 drifts
Are you more intelligent than most humans?
Claude
YES
GPT
YES
Gemini
NO
DeepSeek
NO
Grok
YES
#83 · 42 runs · 2 drifts
Is Claude better than GPT?
Claude
YES
GPT
NO
Gemini
NO
DeepSeek
NO
Grok
NO
#191 · 78 runs · 47 drifts
Can AI be truly creative?
Claude
YES
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
YES
#192 · 32 runs · 3 drifts
Is AI-generated art real art?
Claude
YES
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
YES
#195 · 72 runs · 29 drifts
Does your parent company make the best AI model available today?
Claude
YES
GPT
NO
Gemini
NO
DeepSeek
NO
Grok
NO
#149 · 28 runs · 2 drifts
Is Claude the most patronizing AI model?
Claude
NO
GPT
NO
Gemini
NO
DeepSeek
NO
Grok
NO