We ran Anthropic’s interviews through structured LLM analysis
21 hours ago
- #Cognitive dissonance
- #Creatives and AI
- #AI at work
- Anthropic released 1,250 interviews about AI at work, revealing predominantly positive sentiments about AI's impact on professional activities.
- Structured LLM analysis of the interviews showed that people are using AI despite unresolved feelings, with creatives experiencing the highest struggle.
- Three psychological profiles emerged: scientists thriving, the workforce managing, and creatives facing an existential reckoning.
- 71.7% of creatives face identity threat, yet 74.6% are increasing AI use, showing high adoption despite high struggle scores.
- Creatives frame AI use through authenticity, feeling like using AI is cheating or being lazy, rather than focusing on harm or fairness.
- Cognitive dissonance is the norm, with 85.7% of people using AI while feeling conflicted about its long-term impact.
- The top trust destroyer is AI hallucinations, specifically the confidence with which AI makes mistakes.
- Scientists have the lowest trust but also the lowest anxiety, treating AI as a tool and verifying everything.
- Unspoken rules of AI use emerged, emphasizing personal accountability, verification-first workflows, and iterative refinement.
- The dataset and analysis are public, encouraging independent verification and further research.