LessWrong 2.0 Reader

It seems like such an obviously stupid thing to do that my priors aren't very high (though you're right in that they're slightly higher because it's OpenAI). I think it's telling however that neither Claude nor Gemini shy away from revealing the canary string.

What is the probability they intentionally fine tuned to hide canary contamination?

Seems like an obviously very silly thing to do. But with things like the NDA, my priors on oai being deceptive to their own detriment is not that low.

I'm pretty sure it wouldn't forget the string.

This doesn't guarantee it, you're right. But the obvious way to filter canaried data out is to simply remove any document that has the string inside it. To further filter for articles that only talk about the canary instead of intending to use it seems like a needlessly expensive task, given how few such articles there would be.

Further, given that one use of the canary is to tell whether contamination has happened by checking if the model knows the string, not filtering such articles out would still not be great.

All that said however, I think the fact that GPT-4 had memorized BIG-Bench tasks - which certainly used the canary - means that the contamination could have happened from anything that had the canary in it.

Nope, that's more than enough. Caleb Ditchfield, you are seriously mentally ill, and your delusions are causing you to exhibit a pattern of unethical behavior. This is not a place where you will be able to find help or support with your mental illness. Based on skimming your Twitter history, I believe your mental illness is caused by (or exacerbated by) abusing Adderall.

You have already been banned from Lighthaven. I'm extending the ban to LW too.

After reading all the 2.6 and 3.3 sections again, I think the answer to why the homunculus is attention-grabbing is because it involves "continuous self-surprise" in the same way an animate object (mouse...) is. A surprise that is a present as a proprioceptive signal or felt sense. With PNSE, your brain has learned to predict the internal S(X) mental objects and this signal well enough that the remaining surprisingness of the mental processes would be more like the gears contraption from 3.3.2, where "the surprising feeling that I feel would be explained away by a different ingredient in my intuitive model of the situation—namely, my own unfamiliarity with all the gears inside the contraption and how they fit together." And as such, it is easier to tune out: The mind is doing its usual thing. Process as usual.  

Me: I dunno, probably around 9 pm. [At this point, I’ve merely offered some information; I think most people would not interpret this as an assurance, and would not blame me much if I show up to the party at 8:30 or 10:00 or even skip it altogether.]

Assuming the conversation doesn't delve further into this, if I were your friend I'd actually be very surprised if you didn't show up. The question 'At what time are you going?' assumes that you're going, however uncertain the details. If you wish to convey the idea of 'you might not see me at all' your answer should explicitly include 'but I might not go' because without that clause you're agreeing to attend, at least at some point.

To be clear, I agree with the gist of the piece. I just find it funny how even such a short convo could lead to a quite dramatic misunderstanding.

In any case, as a researcher currently working in this area, I am putting a big bet on moderate badness happening (in that I could be working on something else, and my time has value).

Also, there is counterparty risk if you bet on everyone dying.

(Yeah, yeah, you can bet on something like other peoples belief in the impednding apocalypse going up before it actually happens).

“Rapid takeoff” hypotheses are particularly hard to bet on.

Yes, I am seeing that as well. Technical/philosophical stuff is fine, but the psychology in adult fiction is too complex for an 11-years old to enjoy.

Cell line being immortal doesn't prove that immortal brain is possible any more than microbe strain being immortal.