Davidmanheim comments on How Can Average People Contribute to AI Safety?

[Davidmanheim]

I think we basically agree, but I wanted to add the note of caution. Also, I’m evidently more skeptical of the value of evals, as I don’t see a particularly viable theory of change.