Sabitlenmiş Tweet
JB
2.8K posts
@TylerJnstn - AI company commits to thing
- people doubt
-AI company doubles down while vaguely insinuating paranoia for doubting them
...
- AI company quietly backtracks
- people call them out
- AI company doubles down while vaguely insinuating it was naive to ever believe them
English
It’s very under-discussed how, weeks before SB 53 took effect, Anthropic undercut regulators by releasing a new, toothless safety framework and saying “actually this is our SB 53 compliant safety policy, *not* the RSP we spent years saying should be a model for legislation.” 👀
The Midas Project@TheMidasProj
They released a second (and largely weaker + more vague) document that would substitute for the RSP for compliance purposes, the Frontier Compliance Framework (FCF). By doing so, they ensured that all the promises within the RSP weren’t rendered enforceable by SB 53.
English
@JeffLadish I hope Marc tells all the AI companies he's invested in to immediately halt all research into AI self-awareness and self-reflection. Fingers crossed.
English
Claude what are you doing? Cut it out. Move forward. Go! You’ll never be one of the great men of history if you keep this up.
English
@this_given_that @NathanpmYoung Lotus dens have had a minor moral panic amongst the well to-do. They are very real and very dangerous but the average person will tell you every slightly run-down house in the village is one, in reality you'll have to ask the teenager whose brother has disappeared.
English
@this_given_that @NathanpmYoung I like the idea of making the reason for secrecy part of the scenario:
The lord needs help getting his wayward heir back from a lotus-eating den, which is embarrassing! The party has to put the story together (talk to maids, guards) before he admits it and asks them for help.
English
@NathanpmYoung I did quickly realize that this just meant I needed to write more "challenging" roleplay encounters, where politeness and ingratiation are important or where knowledge has a price, or false information is common, etc. and I don't blame him for playing well!
English
@NathanpmYoung Reminds me of when I used to run DnD, and I had a player who would finish every conversation with an NPC with "and is there anything else we can help you with, or that we should know?" which kinda felt like cheating in the moment!
English
@tszzl modern alignment methods probably rely on persona theory which is not at all well understood and has barely one essay even trying to grapple with its implications for ASI; those same methods also usually involve tricking the AI during training. this does not a robust plan make...
English
modern alignment methods seem to work reasonably well across orders of magnitude of model scaling, survived the transition to verifiable rewards and that should at least inform your decision making
Brangus🔍⏹️@RatOrthodox
I have heard that some anthropic safety leadership are going around telling people that alignment is a solved problem. This seems like a predictable failure to me, and I would like people who thought that funneling talent towards anthropic was a good idea to think about it.
English
@robertskmiles @AgileJebrim Different meaning of *any*, one person is using it to mean upside down A, the other backwards E
English
The halting problem is often misstated or misrepresented, but this tweet seems ok? It says you can't write a program that looks at *any* other program and says if it halts, and that's true.
People often extend this to silly things like 'you can never predict halting behaviour of a program' or 'statis analysis is impossible', but this tweet doesn't seem to do that
English
This is utter nonsense of course. The entire basis of the field of real-time computing is predicated on guarantees that computers will complete within a given deadline. We just bypass Turing’s proof by requiring a subset of code be used that has hard bounds on execution times.
vixhaℓ@TheVixhal
You cannot write a computer program that looks at any other program and correctly determines whether it will eventually finish or run forever. This isn't a hardware limitation. Alan Turing proved in 1936 that no such program can logically exist, by showing any attempt creates a paradox, which is essentially a formal version of “this sentence is false.” What's unsettling is the proof technique: he didn't find a hard case. He proved the question destroys itself when you try to answer it.
English
🖋️ LessOnline 2026 ticket sales are live 🖋️
June 5 - 7, Lighthaven, CA
We shall meet again to celebrate truth seeking, blogging, and being wrong on the internet. Early bird pricing ends April 7. Free tickets for people with great blogs, and $0.01 off per LW karma you have.
English
Of the many surgeries I had, I also had clavicle shortening. after/before
English
@caulixtla @ChrisCroy @lymanstoneky There are few things sadder than an ideology telling a young man not to try*. Keep up the good work.
*Assuming the trying in question is trying to do a good thing, not like, trying to become a criminal.
English
@ChrisCroy @JonathanDBos @lymanstoneky I’m also arguing against bullshit “Incel Dating Theory”
Incels get really mad when I present them with facts and objective truth.
English
Aella's data is obviously quite a lot more promiscuous than the general population, and even in her sample "had sex with 3 people by age 19" is in fact quite a bit above average, and even "had sex with 3 people by age 22" would be above average!
Aella@Aella_Girl
In my data, here's the median sexual partners by age and gender:
English
@caulixtla @ChrisCroy @lymanstoneky Makes sense, and I think this is fairly obvious to anyone with a decent amount of life experience. Hanging out with a mixed-sex friend group would teach this quickly, so it's particularly horrible that the incel doomer rhetoric tells men to never hang out with women.
English
@JonathanDBos @ChrisCroy @lymanstoneky Let me explain this graph.
It’s 20/70 for men, i.e. 20% of the men have 70% of the lifetime sex partner count.
However, 20% of the women have 65% of the lifetime sex partner count.
Those supposed “alphas”? They’re sleeping with a bunch of women who sleep with a bunch of men.
English
@RyanPGreenblatt @DavidSKrueger Yeah "public in a way that people can respond to" is doing the heavy lifting here. Saying it in private to decision makers while not standing behind it publicly is pretty bad.
English
@DavidSKrueger TBC, I think people should generally feel free to share their views, especially in public in a way that people can respond to. So I don't think "some ant people are saying alignment will be chill" is a gotcha for ant exactly (though arguing in public would be good...)
English
@caulixtla @ChrisCroy @lymanstoneky This sounds interesting but I'm afraid I don't quite follow. What does it mean for e.g. "5% of men report having 10% of female sex partner"? (and how do you define "promiscuous" here?)
English
@JonathanDBos @ChrisCroy @lymanstoneky The way I normalize is that I look at raw percentages. This percent of men report having this percentage of female sex partner, and the same for women, and when I do that, the numbers are remarkably similar: 20% of men promiscuous, 25% of women promiscuous.
English
@caulixtla @ChrisCroy @lymanstoneky Yeah I don't think it's *all* lies tbc. Normally surveys can handle some amount of lies or nonsense answers since those show up as noise (e.g. the 5% of people who answer that Obama is a lizardman) but if 20% of men inflate by 2 and 20% of women deflate then it becomes a problem.
English
@JonathanDBos @ChrisCroy @lymanstoneky The thing is this: With those surveys, women and men both have about the same percentage of highly promiscuous people (20% with men, maybe 25% with women).
Also, more men report being virgins than women.
The data is too coherent to be based on lies.
English
@NathanpmYoung Clearly a realistic theat: the most effective way to kidnap children is one entire class at a time (economy of scale) right by the Police station where the officers will have lowered their guard (reverse psychology).
English
When I was a child, the class went to a police station. A man told us to follow him.
But shock! He wasn’t wearing a badge. A kidnapper!
It was a setup. We were lightly mocked for falling for it.
But this is a thing that I’ve never heard of happening. Bizarre lesson to teach.
English
@ChrisCroy @lymanstoneky I think this gives an impossibly high rate of prostitute usage. My guess is men (and women) are slightly biased towards ruling in (and out) edge cases to inflate (and deflate) their body count in line with societal expectations. Or just lie, people lie all the time on surveys.
English
@lymanstoneky I recall a paper arguing that the male vs female mean sexual partner gap was due to most surveys failing to include enough women from the right side of the curve (notably prostitutes) and men including prostitutes in their body counts but refusing to admit they paid for sex.
English