Neural networks don't literally mirror the brain: the name descends from a 1943 math abstraction
vishalmisra · x · 2026-09-26
Responding to the claim that "our brains are literally neural networks," vishalmisra argues the word "literally" is doing heroic work here.
- Brains contain biological neurons; AI neural networks descend from McCulloch & Pitts' 1943 mathematical abstraction, later developed through Rosenblatt's perceptron.
- The name survived, the mechanism did not — inspiration is not identity.
The author shares a side-by-side image of Cajal's biological neurons versus the McCulloch & Pitts abstraction, noting modern neural networks descend from the latter, not the former.
Related event: AI Neural Networks Come From Math, Not Literal Brains(2 posts)→
More from AGI Musings
- Swarm hype critique: a loop is not AGI — frozen models just sampled more — gerardsans · 2026-09-26
- The demise of software engineering is overblown — programming becomes control engineering around LLMs — viksit · 2026-09-26
- emax: teaching "better prompting" has diminishing returns, cost/carbon is a systems problem — emax · 2026-09-26
- David Sacks: AI regulation lobbying could cost Anthropic and OpenAI their 6-12 month frontier lead — victor_explore · 2026-09-26
- Frontier AI Is a Race of Energy, Data, Talent Density and Release Latency — AryHHAry · 2026-09-26
- Is the obsession with 'building' with AI just an addiction? — Dredgefort · 2026-09-26