About this episode

Matt Turck interviews AI researcher Mostafa Dehghani about the future of AI automating its own research and development.

Are we truly on the verge of AI automating its own research and development? In this deep-dive episode of the MAD Podcast, Matt Turck sits down with Mostafa Dehghani, a pioneering AI researcher at Google DeepMind whose work on Universal Transformers and Vision Transformers (ViT) helped lay the groundwork for today's frontier models. Moving past the hype, Mostafa breaks down the actual mechanics of "thinking in loops" and Recursive Self-Improvement (RSI). He explores the critical bottlenecks holding back true AGI—from evaluation limits and formal verification to the brutal math of long-horizon reliability. Mostafa and Matt also discuss the shift from pre-training to post-training, how Gemini's Nano Banana 2 processes pixels and text simultaneously, and why the "frozen" nature of today's models means Continual Learning is the next massive frontier for enterprise AI and data pipelines. (00:00) Intro (01:17) What “loops” in AI actually mean (05:04) Self-improvement as the next chapter of machine learning (07:32) Are Karpathy’s autoresearch agents an early form of AI self-improvement? (08:56) AI building AI: how close are we? (10:02) The biggest bottlenecks: evals…

People in this episode

Host: Matt Turck

Guest: Mostafa Dehghani

Topics covered

• AI research
• Recursive Self-Improvement
• AGI bottlenecks
• machine learning
• Continual Learning
• AI building AI

Keywords

• AI
• self-improvement
• AGI
• evaluation limits
• formal verification
• long-horizon reliability
• machine learning
• data pipelines

Mentioned in this episode

Organizations: Google DeepMind

Products: Universal Transformers, Vision Transformers (ViT), Gemini's Nano Banana 2

More episodes of The MAD Podcast with Matt Turck

• OpenAI Board Member Zico Kolter on the Real Risks of Frontier AI · May 7, 2026 · 1h 17m
• Benedict Evans: OpenAI’s Moat Problem & the Future of Software · March 19, 2026 · 1h 1m
• Everything Gets Rebuilt: The New AI Agent Stack | Harrison Chase, LangChain · March 12, 2026 · 47 min
• AI That Can Prove It’s Right: Verification as the Missing Layer in AI — Carina Hong · February 26, 2026 · 1h 4m
• Voice AI’s Big Moment: Why Everything Is Changing Now (ft. Neil Zeghidour, Gradium AI) · February 19, 2026 · 1h 23m
• Mistral AI vs. Silicon Valley: The Rise of Sovereign AI · February 12, 2026 · 58 min

Explore listener stats, chart rankings, contacts and more on the The MAD Podcast with Matt Turck podcast page.