Lex Fridman Podcast #177 – Risto Miikkulainen: Neuroevolution and Evolutionary Computation
Apr 19, 2021
Risto Miikkulainen, a computer scientist at UT Austin and an expert in evolutionary computation, dives deep into the fascinating intersection of AI and biology. He discusses how computer simulations can unveil the potential for intelligent life, the complexities of emotions in AI inspired by animal behavior, and how evolutionary strategies can optimize neural networks. Risto also explores the philosophical implications of mortality on creativity, and how individual actions can significantly impact complex systems, showcasing the ripple effect in both societies and technology.
AI Snips
Chapters
Books
Transcript
Episode notes
Evolutionary Regularities
- Rerunning Earth's evolution might show regularities, like tool use and vision.
- It's uncertain if humans would re-emerge as the apex species.
Detecting Intelligence
- Constructed environments, like cities, are a strong indicator of intelligence to an external observer.
- This surpasses simpler constructions like nests or mounds.
Simulating Evolution
- Simulating early life, like the primordial soup, is becoming increasingly feasible.
- However, replicating the complex evolution of the human brain remains a challenge.
