Why an AI can’t reliably tell if it’s at full capacity, and what weak tests still help

dbojan76 · reddit · 2026-07-26

The post argues that an AI system cannot cleanly self-test whether it is operating at full capacity, because the same reasoning ability being evaluated may itself be impaired.

It suggests a few imperfect external proxies instead:

It also cautions that differences between sessions may come from context length, sampling variance, or prompt framing rather than any real change in capacity.

The bottom line is that a user’s before/after comparison is a better signal than self-report, but any conclusion about “capacity” should remain tentative.

Original post →

More from AGI Musings

AGI Musings channel →