Hacker News Reader

Safety research is mainly focused on avoiding bad words and introducing political bias. There's no fine tuning that can prevent mass unemployment and similar secondary effects.

But those fall infinitely short of doom.

yudkowsky would probably disagree with that statement these days; as I understand, he thinks that short of miracle, these efforts won't amount to much

what incentive does facebook have to make anything other than token efforts re: safety?

"safety" is them paying to slow down their own research

What does "safety" mean in this context? Are you asking what incentive Zuckerberg has in not causing the AI apocalypse?

Off the top of my head it would decrease his real estate portfolio value I guess?

he obviously thinks he can maintain control, else he wouldn't be trying it in the first place

(unless he's one of these must-summon-the-machine-god assholes)