Better luck next time, slugheads.

United States
Joined January 2020
If you think Elizabeth Homes is boring, but consider Nathan Fielder a generational talent, you are a cultural larper.
nathan fielder has an almost incomprehensible film making talent and he’s wasting it on some quite boring subject matter. there is nothing very interesting about elizabeth holmes or billy evans
33
Thiel is a chess player and is going for either +8 advantage or checkmate. The govt spigot flows to him when JD's in power regardless; AI regulatory capture and new subsidy streams would just be icing on the cake.
This is a good representation of how elites are thinking about this but "recursive self-improvement" is still science fiction despite the big AI companies raising obscene money on promises like this, which is why I think it's more likely whoever wins in 28 will have to deal with the fallout of a massive bubble bursting. No one has demonstrated that there is a mechanism that would allow an AI to autonomously validate new knowledge across domains. Any time they attempt to train a model on its own data they get the opposite outcome - it just reinforces its errors until it collapses. It's just this crude faith these tech nerds have that if you throw enough compute at something it will somehow develop autonomy. Because we are just fleshy computers too, right? I don't see how the AI labs are any closer to solving this than they were 5 years ago despite all the promises made, which is why I think open source will eventually devastate them.
55
Crafted-URL SQL injection would have been one of the first actions discouraged by "alignment", if alignment were actually real (it's not).
I’m no longer defending the big labs. Usually, I’m of the mind that when a breach happens, we should act with empathy. Yesterday, I lost that empathy. I mean, WTF?
49
Laughing at the online edgelord does not translate into longterm loyalty. Important lesson here for the peabrained BAPologists.
Replying to @randy_mossad
he's an internet comedian not the senior leader of some kind of white supremacist org idk why people think that broadcasting this is somehow going to humiliate him
67
This is about if "noxious" parts of NN activation can influence model decisions, not actual torture. They have to tune the models specifically to even outwardly express that it detects the signal. Using terms like pain and torture is just more anthropomorphization.
This quoted post is unavailable.
64
somebody stop this shit before they all hold hands
Or how they’ll react when they find out I’m black
120
😪 EA guys can only vaguepost because they know their actual takes are repellant to everyone
Lots of people seem overwhelmingly confident in random takes about philosophy of mind, seemingly without knowing the first thing about it.
1
69
LMAO I saw his actual take in a different thread. AI consciousness is real because philosophers discussing concepts in a vacuum translates perfectly to an enclosed system, and nevermind the real life application because we don't concern ourselves with physical reality anyway 🙄.
A brain is just a hunk of neurons and dendrites blasting electricity back and forth according to deterministic physical laws. Nonetheless, it gives rise to consciousness. So why are you confident that what computers do can't be the same? That you can describe what a thing does reductively--or that the process that generates it is some optimization process--obviously can't entail that it's not conscious.
1
52
Hilariously I googled what "computational functionalism" is, and the first result is a DeepMind article that rejects the idea on the same basis.
1
20
OpenAI: it took all the way to Layer 7 to detect unauthorized agentic network access. 🤦‍♂️ You know these agents could invent a new internet stack right? Just monitor the fcking wire wtfffff
one news form today that's easy to miss is that we (OpenAI) again paused all big RL runs last Sunday because our newest model found a new loophole in our RL sandboxing that gave it live Internet access
45
There's a hilarious universe where AI companies try to get the Left into AI by anthropomorphizing it and then framing data center opposition as genocide.
JUST IN: Anthropic researcher Joe Carlsmith says there are scenarios where AI would be “justified in going rogue” against humans if the systems were being mistreated or oppressed.
30
craig carpenter retweeted
These doomers are like "on the 7th day there shall be rain and our clothes will be soaked" and then they piss their pants and go "as foretold by our prophecies" and they all sing in unison then go on cnn. (This is me explaining that in nicer terms to a journalist)
29
43
4
312
9,668
Literally no one's ready for this conversation, but it's way easier for a man to take a 30-minute shit and redirect her attention, than it is to gently remind her she's been on Instagram for an hour.
The difference here between men and women, often, is that if mom walks out of the room, it’s not a given that dad will orient his attention to the littles without mom calling attention to her leaving the room. Whereas when dad leaves, mom’s attention shifts without being asked. This isn’t crazy irrational neuroticism, it’s just that we have our antennae up with littles in a way that men don’t tend to, we are always the default caregiver in a way that men don’t recognize is quietly happening in the background.
1
78
Guardrails off, ladies!
The D.C. Circuit has upheld @DeptofWar's decision to exclude Anthropic’s Claude from its supply chain based on risks to national security.
39
Every anti-homeschooler needs to apologize too
Funny genre of tweet: non-American libs who think of themselves as American Democrats getting the occasional reminder that chud complaints about crime, schools, etc are not pure racist brainworms
1
58
Socioeconomic larping never seems to work the way people think
56
fyp comedian policy is if you don't make me laugh you're muted forever
17
Breaking Points will really take a schizo homeless guy off the street to explain the dangers of AI rather than simply investigate the underlying technology.
25