LWiAI Podcast #232 - ChatGPT Ads, Thinking Machines Drama, STEM
Summary
Hosts Andrei Kurenkov and Jeremy Harris discuss AI safety risks like authoritarian capture, OpenAI’s ad rollout and age detection, and Google’s Gemini SAT practice. They cover startup drama at Thinking Machines, China’s progress with Huawei chips, and Samsung’s rise as a key chip manufacturer amid TSMC bottlenecks.
Topics discussed
Introduction and podcast housekeeping
AI safety, authoritarianism, and superintelligence risks
OpenAI tests ads in ChatGPT and new pricing tiers
ChatGPT Go for minors and AI in education
AI-powered SAT prep and Chinese chatbot competition
Thinking Machines drama and exodus to OpenAI
China's AI chips, TSMC bottlenecks, and foundry wars
Elon Musk's XAI gigawatt cluster and power strategy
New AI startups and fundraising rounds
Open source models, video datasets, and agent benchmarks
Technical deep dive: Efficient transformer architectures
Reasoning models and societies of thought
AI for science: Limitations and hallucinations
Deepfake legislation and safety probe research
Anthropic releases Claude's constitution
Artists vs AI, cultural backlash, and conclusion
Listen ad-free on Castria