The Natural Abstractions Hypothesis

Created time
Mar 17, 2023 04:35 PM
Main Box
Tags
Natural Abstractions
Interpretability
AI Alignment
Public

The Natural Abstractions Hypothesis

The NAH states that our universe abstracts well. This means that there are several high-level summaries of low-level systems.
Furthermore, these summaries are natural, in the sense that we expect a wide range of intelligent agents to converge on them.

Source: Chan, L., Lang, L. and Jenner, E. (no date) ‘Natural Abstractions: Key claims, Theorems, and Critiques’. Available at: https://www.alignmentforum.org/posts/gvzW46Z3BsaZsLc25/natural-abstractions-key-claims-theorems-and-critiques-1 (Accessed: 19 March 2023).