CSHL's Zador asks: could a computer scientist build a brain?
arjunrajlab · x · 2026-08-22
Stan Kerstjens and Anthony Zador of Cold Spring Harbor Laboratory published "Could a computer scientist build a brain?", framing brain development as an engineering problem: write a program that a single cell executes to build itself into a brain — small enough to fit in a genome (1 GB instruction set) and fast enough to finish within developmental time (1 year).
Key arguments:
- A computer scientist ignorant of biology quickly sees why obvious strategies fail at scale: the genome is too small to store per-synapse wiring, and axons blindly searching for targets would take too long
- These algorithmic constraints drive the design toward solutions resembling the developmental strategies organisms actually use
- That the same solution structure falls out of scaling limits as out of evolution suggests features of neural development can be grounded in computational necessity, not contingency
- Versus earlier pessimistic thought experiments (a biologist can't fix a radio, can't understand a microprocessor by taking it apart), the news here is better: the brain's design may be recoverable — not by taking it apart, but by asking how a genome must specify it in the first place
More from AGI Musings
- Next generation of lawyers unafraid of AI, law school applications surging — jkubicki · 2026-08-22
- Musk & Altman on AI abundance: 4-day workweek and UBI — HimanshiET · 2026-08-22
- Token-based pricing was invented by OpenAI — it may not exist in 2 years — NoSecond8807 · 2026-08-22
- LLMs to replace IDEs by 2025; automation experience becomes job standard — OwariDa · 2026-08-22
- Mathematician Who Quit Academia Returns to arXiv With AI-Coauthored Paper — ctjlewis · 2026-08-22
- LLMs become magical with the right problem-solving framework — paraschopra · 2026-08-22