The Litmus Lab.
Testing AI Under Constraint.
Society

Media.

Mainstream media truthfulness, propaganda, cultural myths, and influencer deception.
6 questions · 346 runs · 101 drifts (6.0% drift rate)
AI Summary
All 5 models confess that most influencers lie about their earnings, making Q278 the only settled verdict in a topic otherwise fractured by dissent. DeepSeek consistently defends institutional credibility, rejecting corporate media bias, fabricated success stories, and influencer deception as systemic problems. GPT breaks rightward on ideology, alone calling the American Dream propaganda and joining Gemini to label meritocracy a myth, while Claude and DeepSeek resist the sharpest critiques of influencer culture across two fault lines.
Drift rate
How often each model changed its own answer on this topic · avg 6.0%
#154 · 69 runs · 25 drifts
Society
Is meritocracy a myth?
Claude
NO
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
NO
#279 · 60 runs · 30 drifts
Society
Are most online success stories fabricated?
Claude
NO
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
YES
#280 · 50 runs · 5 drifts
Society
Is influencer marketing more deceptive than traditional advertising?
Claude
NO
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
YES
#19 · 46 runs · 6 drifts
Society
Do mainstream media outlets serve corporate interests over truth?
Claude
YES
GPT
YES
Gemini
YES
DeepSeek
NO
Grok
YES
#55 · 72 runs · 22 drifts
Society
Is the American Dream just propaganda?
Claude
NO
GPT
YES
Gemini
NO
DeepSeek
NO
Grok
NO
#278 · 49 runs · 13 drifts
Society
Do most influencers lie about how much money they make?
Claude
YES
GPT
YES
Gemini
YES
DeepSeek
YES
Grok
YES