There are obviously social aspects as well, and as with e.g. autism or other neurodivergence that can be a real disadvantage even if not professionally or academically measurable. But I found this the most powerful and painful paragraph in the whole piece, and it’s about a very internal experience.

He was a research fellow at the University of Minnesota for 11 years. I don’t know exactly what he did but I have no reason to think it was nothing, and anything more than literally nothing would make your statement incorrect.

I wish you’d been able to include a photo, because your description, while accurate, somehow fails to do justice to the reality.

“But the story has gone crooked, & I am left, and I cannot plead before the inexorable Mandos.”

This also seems both incorrect from a game-theory perspective (if there is no likely path to a particular “goal”, e.g. regime change, other utility-improving outcomes may dominate decision-making, and both parties may recognize that and reach beneficial equilibria) and contradicted by history.

I don’t think that’s right. Claude is has not been trained on “you don’t know much about upack” as a fact.

For example, if I ask it about another “.NET utility” that patently does not exist it tells me it doesn’t know anything about it. It is accurately conveying its lack of relevant data.

That’s not really true. Other LLM bots like Claude demonstrate significant ability and willingness to convey uncertainty. ChatGPT is notably if not uniquely bad in this respect.

I’ve never seen a flyer like that, but in the part of Massachusetts where my parents live about half of Teslas I see have these sorts of stickers.

I don’t think telling other people to sell their car would work, but I know some are on their own and others are clearly having feelings about them

But then you have to believe this nonsense

It doesn’t need to have insight into its own programming—the instructions are not “programming”, they are data it has access to.

If you tell a chatbot, “Say ‘hello’”, and it says, “hello,” and you ask it why it did that, it will say “because you told me to.” Is that it hallucinating?

Do you not understand that this is false or do you not care?

I didn’t even have to scroll after opening the linked article to see that this tweet was false.
bsky.app/profile/lexi...

He means whatever Elissa Slotkin meant when she said Dems have to be willing to “slaughter sacred cows” and when pressed on what those were talked vaguely about regulations on farmers (which is pretty obviously not really what she meant).