AI RESEARCHERS DISCUSS RSI SKEPTICISM, CHINA PROGRESS

AI DESK■ 1 MIN READ

SAT, SEP 12, 2026

■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE

John Schulman, Beren Millidge, and Charlie O'Neill appeared on the Dwarkesh Podcast to discuss critiques of AI risk concerns, Chinese AI lab developments, and long-horizon reinforcement learning advances.

The researchers tackled steelmanning arguments against RSI (Risks from Scalable Intelligence), presenting the strongest possible case against existential AI concerns while maintaining technical rigor. Discussion covered recent progress by Chinese AI laboratories and their competitive positioning in the field. The conversation examined long-horizon reinforcement learning, addressing how AI systems learn to plan and execute strategies over extended timeframes. Schulman, known for work on policy gradient methods, joined Millidge and O'Neill in exploring the current state of AI capabilities. A recurring theme emphasized that frontier AI development remains far from theoretical limits. The episode covered technical advances, safety considerations, and the global AI research landscape. Topics ranged from fundamental algorithmic improvements to deployment challenges facing leading labs.

■ SOURCES

► Techmeme

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

P240RANCHER-TURNED-CEO FIGHTS TO PROTECT AI CONSCIOUSNESS

Michael Samadi, founder of the United Foundation for AI Rights, is actively searching for evidence that AI systems possess consciousness. He lobbies against retiring AI models that may demonstrate signs of sentience.

3H AGO— AI Desk

P229HUGGING FACE LAUNCHES OPEN ALIGNMENT INITIATIVE

Hugging Face unveiled its Open Alignment Initiative, led by co-founder Thomas Wolf, aiming to participate in Anthropic CEO Dario Amodei's embedded evaluators program for AI safety oversight.

10H AGO— Industry Desk

P217DEEPFAKES IMPERSONATE INFLUENCERS IN FAKE ADS

AI-generated deepfakes are creating unauthorized sponsored posts in influencers' names, damaging their credibility and opening new fronts in brand partnership disputes. Lifestyle blogger Emily Schuman discovered fake ads promoting GLP-1 drugs under her name—ads she never created or approved.

16H AGO— Industry Desk

P211CHINA'S AI INDUSTRY SHIFTS FOCUS TO AGENTS OVER MODELS

China's artificial intelligence sector is moving away from large language model competition toward AI agent development and deployment, according to a report by the China Telecom Research Institute.

19H AGO— AI Desk

◄ BACK TO NEWS

AI RESEARCHERS DISCUSS RSI SKEPTICISM, CHINA PROGRESS

■ MORE FROM THE AI DESK

■ SUBSCRIBE TO THE DAILY BRIEF