Rob Bensinger ⏹️

@robbensinger

@MIRIBerkeley. If we build opaque superhuman AIs, it's likely to get us killed. The least-bad solution is an international agreement like intelligence.org/ban.

Berkeley, California

Joined November 2008

Replies Reposts Media Articles

Pinned
Rob Bensinger ⏹️
@robbensinger
Aug 8
I think one of the biggest reasons the world is currently sleepwalking into getting ourselves and our families killed by ASI development is that we're not self-aware about why we're doing that. We can see the smarter-than-human AI disaster approaching, but it's a bit foggy why
Nate Soares ⏹️
@So8res
Aug 7
Replying to @robbensinger
Yeah, practically zero respect for the problem. I don't know what went so culturally wrong. Anthropic is heavily culturally EA, and the EAs kept pretending to be serious people who engaged in serious ways. Why do they fall so far below the bar of serious engineers of yore?
41
Rob Bensinger ⏹️
@robbensinger
Sep 7
Or you can do the same thing but twenty years earlier, through the mysterious and rare art of Thinking Ahead At All.
will depue
@willdepue
Sep 7
”If in 2018 you are not an accelerationist, you have no heart. If in 2026 you are not safetyist you have no brain.”
3
Rob Bensinger ⏹️
@robbensinger
Sep 6
x.com/noshpesoj/stat…
Brangus🔍⏹️
@RatOrthodox
Aug 31
i am completely open w my gf just like oai is completely open w third party evaluators. she can look at my dms as long as she doesn't look at anything before june 25 of this year, or ask any questions to the girl i sent 95% of my dms to. just out of scope for this investigation
1
Rob Bensinger ⏹️
@robbensinger
Sep 6
I think posts like this are downstream of not understanding the prediction, and therefore treating it like a black box that emits sentences. Then you find it suspicious when the box doesn't make strong claims about all kinds of random topics that are nearer in time than ASI. But
corsaren
@corsaren
Sep 5
My problem with this view is that if a failed AI attack attempt is just evidence that the agents aren’t advanced/ASI enough, then the claim that “ASI would never attack humanity until it was sure it would succeed” is predictively empty. Moreover, if witnessing such a failed
1
Rob Bensinger ⏹️
@robbensinger
Sep 4
From the people who brought you "really, aren't corporations basically superintelligences already?" and "really, isn't scientific innovation basically recursive self-improvement already?":
ExileAI
@ExileAI_0
Sep 4
Replying to @gcolbourn
You have swarms of rogue people, government, institutions etc that have done immense harm already. Where was your outrage? . Go ahead and pause everything. I don’t think you will like your vision of a safe utopia.
1