Observed arrival · 2026-10-22
NAI: a search for the law in your data
NAI is a CPU-based symbolic-regression tool that tries to infer a formula from tabular data, then reports how it performs on unseen rows.
- For
- Researchers and analysts with tabular measurements
- Worth noticing
- NAI reports first-place results across four benchmark sets, including 1st of 28 systems on SRBench 2025.
Field notes
The workflow treats a dataset as a table, with measurement error supplied in a sigma column where relevant, then evaluates the discovered expression on held-out rows. The page demonstrates this with examples including plasma concentration, cutting-tool life, and model-training loss; forecast bands and suggestions for the next measurements appear as downstream uses. Its benchmark page is linked from the landing page, which says its rankings are against results published for other systems. The medical illustration is explicitly synthetic and not intended for diagnosis or treatment decisions.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue