Intel / podcast

Dwarkesh Podcast: Adam Marblestone on the Brain

Source: dwarkesh.com
Author: Dwarkesh Patel
Date: Published Dec 30, 2025

Why it matters

A neuroscientist's argument that reward-function design matters more than model architecture gives builders a useful frame for reasoning about RL objectives and strategies, drawing concrete parallels between brain mechanisms and current AI approaches.

Terms in this piece · Glossary

inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.

Key quotes

“I think evolution may have built a lot of complexity into the loss functions actually, many different loss functions for different areas turned on at different stages of development.”
Adam Marblestone

“The fact that we don’t have value functions at all in the LLMs is crazy. I think because Ilya said it, I can say it.”
Adam Marblestone

“The cortex doesn’t know about spiders, it just knows about layers.”
Adam Marblestone

“If every iPhone was also a brain scanner, you would not have this problem and we would be training AI with the brain signals.”
Adam Marblestone

“What I actually want to do is go and map the entire mouse brain and figure this out comprehensively and make neuroscience a ground-truth science.”
Adam Marblestone

Chapters

00:00:00The brain's secret sauce is reward functions, not architecture
00:22:20Amortized inference and what the genome stores
00:42:42Model-based vs model-free RL in the brain
00:50:31Is biological hardware a limitation or advantage?
01:03:59Why a map of the human brain matters
01:23:28The value of automating math
01:38:18Architecture of the brain

Read the source www.dwarkesh.com