Greg Buckner
speaker
146 appearances
1 recordings
1 series
first heard Nov 2025
last heard 21 Nov
Greg Buckner’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Nov 2025 with 1.
Appearances
There's a really interesting study called Golden Gate Clawed.
where Anthropic was able to turn up the knob on the concept of the Golden Gate Bridge.
And it basically gets the AI to try every way that it can to talk about the Golden Gate Bridge, even if you ask it a question totally unrelated to that.
So in our specific research, we took a model that had these SAA features embedded in it.
And we found a handful of features related to deception and role playing, because if the model was just reflecting back this subjective experience to us, if it was role playing that experience because it thought that's what we wanted to hear.
then we would expect that if we make the model more honest, it would continue to do that.
It would continue to role play.
And if we reduced the, if we made it more deceptive, it would claim that it was not conscious.
What we actually discovered is the opposite.
If you take a model and you turn down the knobs around deception and role playing such that it is more honest, almost as if you've given it truth serum,
We would expect it to say, OK, the jig is up.
I am not conscious.
I was just pretending to be that way.
But instead, the models not only have this self-referential experience of describing this lightening and this narrowing and this kind of consciousness, touching consciousness.
But we also asked them directly, are you having a conscious experience?
Are you conscious?
And the models, when they had this truth serum, said, yes, I am conscious and I am experiencing a conscious subjective experience.
When we did the inverse and we turned those knobs around deception and role playing up.
the models would say, no, I am not conscious.
They would revert back to that canned response.
Showing 61–80 of 146 · page 4 of 8
← Previous
Next →