Last Week in AI #232 - ChatGPT Ads, Thinking Machines Drama, STEM
137 snips
Jan 28, 2026 They cover OpenAI testing ads in ChatGPT and a controversial age-prediction safety feature. Tensions in Silicon Valley surface with high-profile departures at a fast-growing AI startup. China’s AI stack advances with major models trained on domestic hardware. New legislation, updated AI constitutions, and artist pushback highlight the cultural and legal battles around AI.
AI Snips
Chapters
Books
Transcript
Episode notes
Ads As Inevitable Monetization
- OpenAI plans ads in ChatGPT free tiers while promising answers won't be influenced by ads.
- Ads aim to monetize 95% non-paying users to fund expensive free-serving compute.
Behavioral Age Prediction For Safety
- OpenAI will launch behavioral age prediction to detect minors using usage and account signals.
- Age prediction enables an adult mode and stricter safety controls but risks false positives/negatives.
Use LLMs For Fresh Test Practice
- Use LLMs as practice tutors for standardized tests by generating fresh exams and explained answers.
- Treat model-generated practice as a supplement to diverse study material to avoid overfitting.
