x
The E-Coli Test for AI Alignment — LessWrong