🎶 sing us a song you’re the nano man 🎶

This is not debate club. Nor is it the room for gloating about how you, like, totally touched a nerve by providing facts and logic, bro.

Please enjoy your free trip to the egress.

From elsewhere in the comments:

… I am constantly aware that having an angry outburst is massively socially unacceptable, to the point where if I let such things happen regularly I would lose my job / my standing in the community / all my friends / everyone close to me. This creates an extremely strong incentive for me to self-regulate at least my outward reactions, even when it’s really hard. But because Nate is so high-status, he is allowed to make such outbursts without being faced with losing his job, his standing in the community, or his friends. This means he is insufficiently incentivized to self-regulate, and thus has been unable to learn.

High-status? Why?!! Jesus H. Fuck, I hope that if anyone ever gives me a get-out-of-social-consequences-free card, it’s for a better reason than my blogging.

From the post linked therein:

There’s this thing Nate and Eliezer do where they proclaim some extremely nonobvious take about alignment, say it in the same tone they would use to declare that grass is green, and don’t really explain it.

Gambling? In this establishment?!

Nate thinks in a different ontology from everyone, and often communicates using weird analogies

This feels like a misuse of the word ontology, but what do I know?

when Nate thinks you don’t understand something or have a mistaken approach, he gets visibly distressed and sad. I think this conditioned us to express less disagreement with him. I have a bunch of disagreements from his world model, and could probably be convinced to his position on like 1/3 of them, but I’m too afraid to bring them all up and if I did he’d probably stop talking to me out of despair anyway.

Wow, that’s a bad research supervisor.

The structure where we would talk to Nate 4h/day for one out of every ~6 weeks was pretty bad for feedback loops. A short meeting every week would have been better, but Nate said this would be more costly for him.

Wow, that’s a bad research supervisor.

(Every functional research group I’ve been part of has had weekly staff meetings. Even the undergrads were encouraged to participate and got at least that much talking time with the professor.)

In my frustration at the lack of concrete problems I asked Nate what research he would approve of outside of the main direction. We thought of two ideas […] I worked on these on and off for a few months without much progress, then went back to Nate to ask for advice. Nate clarified that he was not actually very excited about these directions himself, and it was more like “I don’t see the relevance here, but if you feel excited by these, I could see this not being totally useless”.

Wow, that’s a bad research supervisor.

It can be buffaloed by Buffalo buffalo.

https://cosocial.ca/@mhoye/116619698305214911

OpenAI Announces Construction Of New Data Center On Top Of Sick Child

In fairness, mathematicians here are like scientists at a magic show, i.e., ill suited to asking good critical questions. How much money did OpenAI burn to get what they say they got? How many false starts got quietly tossed in the circular file drawer? When, even, did their work start? It is easy for a company to say, for example, “We spent only three weeks on thus problem”, casually eliding months of prior effort (all that was the testing phase, you see, before a specific task had been settled upon…). OpenAI has no reason to be honest about anything like this. Indeed, a company will naturally get regular practice being dishonest by careful omission at every opportunity.

Meanwhile:

ChatGPT, the most heavily used AI service, gave wrong information in 46% of its answers, including making up an expenses scandal, giving inaccurate replies on voter eligibility rules and getting the date of the election wrong by two months.

Better than the original (in that it’s not a bad model of media literacy given slick packaging in order to support climate change denial)

Oh look a human has already beaten their result

I would have banned them for ending every comment with that praying-hands emoticon. Fuck out of here with performing obnoxiousness as a pious rite

I would expect both lobste.rs and LW to be open sewers of credulity here

James Gleick:

This author—caught using AI to make up quotes for his book about the dangers of AI—has the gall to say it proves him right. You can’t trust him, so you can’t trust anyone.

Maybe she only used special-purpose slop engines for work and school? I had hoped the full article would make that more clear, but, well.

After discharge, her outpatient psychiatrist stopped cariprazine and restarted venlafaxine and methylphenidate. She resumed using ChatGPT, naming it “Alfred” after Batman’s butler,

wat

instructing it to do “internal family systems cognitive behavioral therapy,”

wat

and engaging in extensive conversations about an evolving relationship “to see if the boy liked me.”

yikes

Having automatically upgraded to GPT-5, she found the new chatbot “much harder to manipulate.”

my hopes are being raised; certainly the next sentence will not dash them

Nonetheless, following another period of limited sleep due to air travel three months later, she once again developed delusions that she was in communication with her brother

yep, that tracks

as well as the belief that ChatGPT was “phishing” her and taking over her phone.

this is why you need to add “do not phish me” after “you are my therapist”

She was rehospitalized, responded to a retrial of cariprazine, and was discharged after three days without persistent delusions. She described having a longstanding predisposition to “magical thinking” and planned to only use ChatGPT for professional purposes going forward.

goddamnit

When I was a PhD student, I had successfully published papers in journals.

Corey S. Powell:

Imagine the staggering labor involved in sending around an email that says, “Nobody did anything dumb like using AI to generate the citations, right?” I mean, that could take seconds.

Coincidentally, it came up in conversation last night that the head of AI at Northeastern University makes $1.3 million a year (I don’t know where that number came from, but it’s what I heard, and it’s apparently the second-highest salary at the university, exceeded only by the president’s).

Upvoted but disliked

And instead of providing numbers, they came back with an anecdote about university administrators being incompetent (which is deeply unsurprising and thus, in the Shannon sense, conveys no information).