Factually! with Adam Conover An AI Expert Challenges an AI Skeptic, with Ethan Mollick
154 snips
Dec 31, 2025 
In this intriguing discussion, Ethan Mollick, a Wharton School professor and AI expert, challenges skepticism by highlighting AI's practical benefits across fields like education and diagnostics. He delves into the nature of AI hallucinations and clarifies how models can improve tasks. Ethan emphasizes the importance of responsible AI policy, arguing that AI will transform jobs rather than replace them entirely. He offers hopes for democratizing technology while acknowledging the risks of deepfakes and misuse. Their dynamic conversation touches on the future landscape of AI in daily life.
AI Snips
Chapters
Transcript
Episode notes
Hallucination From A Simple Trivia Query
- Adam asked ChatGPT whether Bruno Kirby was in The Godfather Part II and received a confidently wrong answer.
- That hallucination made Adam question using AI for simple factual queries going forward.
Hallucinations Often Come From Model Routing
- Ethan explains hallucinations arise from model routing and design, not magic.
- Cheaper/chatty model routing (e.g., via Siri) often yields faster but less accurate answers.
Use Advanced Models For High-Stakes Tasks
- Use AI as an expert advisor with caveats: it can boost diagnostics, tutoring, coding, and research.
- Prefer recent, higher-capacity models and supervised workflows for critical tasks.