Rob Wiblin
speaker
1,787 appearances
3 recordings
1 series
first heard Sep 2024
last heard May 2025
Rob Wiblin’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
So so you're saying you you might end up with another model that is even more powerful, but it like thinks in new release, we can't understand what thoughts it's having.
But we could make like w so but at the same time we want to have another model that only thinks in chain of thought, it's not able to to sneak anything past us.
It's gonna be worse because it has this handicap that it has to think in into in in in English or in yes,
yeah.
previously had.
So we're still improving just not as much as we might have otherwise, but we've gained interpretability.
Yeah.
Are you trying to grow the grow the team a lot or, you know, uh have a smaller number of people who, you know, are are extremely committed to the mission and maybe like, you know, ha have very productive relationships with with one another?
Yeah.
You could potentially try to automate a lot of the work that you're doing at some point, right?
And and then and at that stage where you're using AIs to do this, I guess possibly in a couple of years' time, then it's more you get like linear returns of just having more money because it means more compute, more thought going into it.
Yeah, okay, makes sense.
Um, I mean, I I is this something that you might turn your attention to is thinking, well, we have this whole research agenda, but we as poor human beings are not uh are not gonna be able to solve all of these questions uh by by by the deadline.
So instead we are gonna just start writing briefs that we can hand off to to um to AIs, you know, a couple of months before the intelligence explosion that we're gonna like dump a whole lot of compute on it with with all of the money that we saved.
That's the plan.
Yeah.
Are there any particular lines of research that you would love to see people do outside of Redwood, maybe 'cause you know you don't have capacity to cover all of it or it's just it's not your main focus?
I mean there's a bunch of
Isn't some of the scalable oversight work happening inside Anthropic?
Is that part of their agenda?
Showing 521–540 of 1,787 · page 27 of 90
← Previous
Next →