Ethan Mollick shared an AI thinking trace from an experiment using two different AI models, GLM-5.2 and Opus 4.8. The experiment involved asking the models to suggest two poems that best represent the current state of Generative AI, with a specific instruction not to simply pick popular poems and justify them afterward, but to think critically about the options first. Mollick noted that the two models exhibited distinct personalities during this interactive process. AI
IMPACT Provides insight into the differing 'personalities' and reasoning processes of various AI models.
RANK_REASON The cluster describes an experiment and observation by an individual about AI model behavior, fitting the commentary bucket.
Read on Bluesky Jetstream — AI desk →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
