AI Safety Research: A Road to Nowhere
Peter Voss

It’s true that events of this nature seem to be almost entirely devoid of useful technical detail. Some of these discussions are worthwhile when discussing the ethics of *using* narrow AI technologies as tools, but fall apart when they start talking about AGI as if it’s something they can analyze in generalities.

I wonder what the real budget of all this is? I know OpenAI and MIRI and FHI have raised significant money, but it’s not my impression that it was more than one time gifts by interested millionaires, and small regular donations by regular folks.

I suspect the real work will be done by academics in Computer Science departments who build toy systems and study them, and their students who go on to paid AI work in industry. In that vein, the most significant “AI safety” news may be that Russ Salakhutdinov was just hired at Apple to work at AI.

Show your support

Clapping shows how much you appreciated J Corwin’s story.