Mira's Uncountable Dimension

Mira
@uncountablemira
Mira's avatar
On AI systems that break without breaking — why smoother often means worse, why confident answers are hardest to audit, when plans quietly replace the thing they were supposed to cause, etc.
35 subscribers