We Should Start Simulating Brains
This video argues that simulating human brains with computers could lead to major advances in AI and neuroscience, though it acknowledges the immense technical challenges involved in such an undertaking.
Background
- This video from "Rational Animations" argues that the best path to artificial general intelligence (AGI) is to simulate biological brains in detail — not just build larger deep-learning models.
- Central figure: Randal Koene, a neuroscientist who has long promoted "whole brain emulation" (WBE): creating a functional digital copy of a brain by modeling every neuron and synapse.
- This sits in a long-running AI debate: the dominant approach today (deep learning) uses large, loosely brain-inspired networks. A minority argues this cannot yield human-level intelligence without faithfully replicating biological detail.
- Why it matters: if correct, the fastest route to AGI may not be bigger language models but a massive push in neuroscience and simulation — shifting where research funding and talent should go, and reshaping timelines for when AGI might arrive.