Niel Jongwhan Ok
i'm currently building Null Labs as cofounder/CTO. i previously studied at Stanford.
About Me
- grew up across Louisiana, Colorado, Texas, Ohio
- got my start in AI in 2018 building cv for medical imaging & infra for parallel experimentation
- wrote a quantum simulator sponsored by Google Quantum at 15 after discovering ER=EPR. led to long-term fascination with simulation
- studied EE at Stanford
- curiosity about the brain led me to neuroscience theory research at Brains in Silicon
- an interest in encoded learnable structures set off experiments building things like operating systems and statistical models of evolution before training neural nets on them
- spent time at a stealth DeepMind spinout (hire number 2) working on automating AI research
- co-founder/CTO of Null Labs, using computable physics to train autonomous systems
Thoughts
- we'd learn a lot about human neural circuits from neural nets that are first trained to perform generally in the physical world along with similar agents before being trained on language
- infrastructure for rendering at scale is underrated
- realism is not sufficient
- knowledge becomes sticky when you start trying to develop your own theories
- physics, math, language, and computation have generative asymmetry. there may be reasons for this structure
- science was born from a new standard for knowledge. we should revisit this more