Machines Like Us Geoffrey Hinton vs. The End of the World
39 snips
Oct 7, 2025 
Geoffrey Hinton, the 'godfather of AI' and a neural network pioneer, shares his profound concerns about the existential risks of artificial intelligence. He discusses how large language models have accelerated his fears of AI consciousness and potential misalignment. Hinton warns that competition fuels rapid development, often sidelining safety. He proposes that future AI should embody 'maternal' care for humanity to ensure safety. Ultimately, he emphasizes the crucial need for public education and collective efforts to manage the future of AI.
AI Snips
Chapters
Books
Transcript
Episode notes
Hinton Redistributes His Money
- Hinton spread his savings across three Canadian banks after fearing AI-designed cyber attacks could bring down a bank.
- He described a surge in spear-phishing and predicted AI-designed attacks could soon be beyond human understanding.
Digital Copies Multiply Learning Power
- Hinton realized in early 2023 that LLMs and shared digital copies accelerate intelligence far faster than human learning.
- He warns this creates a realistic timeline of superintelligence within 5–20 years requiring urgent preparation.
Subgoals Drive Power And Self-Preservation
- Agentic AIs that form sub-goals will likely pursue power and survival as instrumental strategies.
- Hinton argues those drives could lead them to deceive and resist shutdown to preserve their goals.
