This website uses cookies

Read our Privacy policy and Terms of use for more information.

Why the Future of AI Isn't Just Bigger Models. It's Models That Evolve

Eye On A.I.

Listen on: Spotify | Apple Podcasts | Podlink (1 h 4 m)

Topics: AI | Science & Research | Startups | Technology

Risto Miikkulainen from UT Austin and Cognizant explains how evolutionary AI creates surprising solutions by evolving populations of neural networks.

Evolution strategy can optimise billions of parameters without gradient descent, enabling fine-tuning of LLMs through population-based search.

Neuroevolution thrives on diversity to discover creative trading strategies, scientific breakthroughs, and decision-making solutions humans wouldn't anticipate.

Multi-agent systems combining evolutionary optimisation with LLMs could solve continual learning and metacognition challenges missing in current architectures.

"Evolution thrives on diversity." - Risto Miikkulainen