Not only are we
unable to predict which specific AI capabilities will emerge at
higher levels of scale, but we also lack the means by which to assess
an existing AI system’s full range of capabilities.
It can take
years of public experimentation for users to fully probe an existing
AI system’s entire capability surface, and even then the view may
still be incomplete. Indeed, it has been clear since GPT-3 that
scaled AI models possess significant latent capabilities that often
are not recognized until long after their development and public
release.
- Gladstone AI Survey, pg. 24
No comments:
Post a Comment