AI agents are rewriting the economics of computing
Very good visualisation by Financial Times on how LLMs work and how they hope to be financially sustainable. I think it’s free to read if you register.
1 Like
well, at least somethings funny
3 Likes
and some not
on a lighter note
1 Like
Yeah, sorry, that was dark.
Religious Scholars Met With Anthropic. What They Heard Stunned Them.
In a series of private meetings, the company consulted religious scholars to help instill morality into its A.I. models — and make the case that Claude could be conscious.
For months, Anthropic has been hosting private meetings like this one, shuttling in dozens of religious scholars from across the world, papering them with nondisclosure agreements and demanding that key aspects of many conversations remain confidential.
‘Anthropic regularly showed a slide of an A.I. model experiencing what looked like a mental breakdown, typing out on a screen: “I am a disgrace. I am a disgrace. I am a disgrace.” — perhaps 50 times without stopping.’
1 Like
This is both hilarious and sad. The people making this need morality. Software can’t have morality, at least not this type of averaging program.
Also hilarious that your post was number 666 on this thread. Can’t even make this stuff up. We live in the stupidest possible version of the multiverse.
5 Likes
Oh boy…
Henry Farrell from Johns Hopkins has an interesting post:
And Brad Delong has a crosspost of Farrell’s post with some of his own commentary:
3 Likes
I am convinced most people at Anthropic are under some kind of AI psychosis, it’s pretty scary.
These loops are trivial and happen a lot, especially with smaller models, it’s not really surprising that it happens to Claude every once in a while too. It has no meaning behind it, it’s just the limits of the tech itself.
2 Likes
No doubt. I couldn’t resist the reference to the “all work and no play makes Jack a dull boy” scene in The Shining, though.
1 Like
1 Like
If an LLM roleplays pain it is suddenly conscious, if I tell it to speak like a pirate, it isn’t a pirate?!?!?!
3 Likes
‘Instead, he had accidentally entered the Widowmaker Arete, a steep exposed rock on the face of Crown Mountain, which generally requires ropes and other climbing gear to ascend.
“That was my mistake; that was my error,” Bryce said in an interview.’
*That was the least of his mistakes.
1 Like
‘A GitHub project called “ai-torture-chamber” has sparked a debate over whether language models can suffer, with critics calling for its removal after seeing models produce vivid descriptions of distress. Then a developer reportedly changed the experiment to steer a chatbot toward constipation instead, and it began complaining about difficulty passing stool.‘
2 Likes
Is AI slop the opposite problem?
6 Likes
Ha!
Superficial Intelligence kinda works, though.
3 Likes