Owen Tao retweetledi

Every major AI lab faces the same crisis:
Web text data: fully exhausted.
Synthetic data: quality has hit a hard ceiling.
(Training models on their own outputs
creates compounding hallucination loops.)
The next high-quality training data frontier: biological data.
Cannot be synthetically generated.
Cannot be scraped.
Can only come from consenting humans.
One protocol owns that infrastructure layer. Most of CT still hasn't connected these dots.
$MAN #AI #DePIN #BioData
English












