Super Data Science: ML & AI Podcast with Jon Krohn 919: Hopes and Fears of AGI, with All-Time Bestselling ML Author Aurélien Géron
92 snips
Sep 2, 2025 
Aurélien Géron, author of 'Hands-On Machine Learning,' shares his journey and insights into the fourth edition of his bestselling book. He discusses the pivotal shift from TensorFlow to PyTorch, emphasizing hands-on learning for innovation. Aurélien expresses both hopes and fears regarding AGI, addressing ethical dilemmas and alignment challenges. He also highlights the urgency of aligning AI with human values, reflecting on its potential transformative role in education and society. This engaging conversation balances optimism with caution in AI's rapidly evolving landscape.
AI Snips
Chapters
Books
Transcript
Episode notes
Keep Foundational Papers, Prune Details
- Field trends evolve: GANs were dominant until diffusion models improved image quality and stability.
- Keep foundational concepts and exciting topics even if some techniques shrink in scope or move online.
Don’t Skip Reading The Code
- Keep coding knowledge: LLMs help but they still miss subtle bugs like a missing assignment line.
- Read and step through code yourself because it often makes concepts click better than diagrams alone.
AGI Timeline And The World-Model Gap
- Aurélien expects AGI within roughly 5–10 years but notes current LLMs lack deep unified representations.
- He sees promise in higher-level world models (e.g., JEPA) that predict abstractions rather than pixels.
