lukeprog comments on CFAR’s new focus, and AI Safety

[lukeprog]

“Accuracy-boosting” or “raising accuracy”?