←
ueaj9,388 posts
ueaj avatar

ueaj

@_ueaj

Researcher - ueaj.dev/ prev: @pangram @blocks

◷ Joined January 2025▧ 680 Photos and videos
ueaj avatar
ueaj@_ueaj·17 Nov 2025

guy who spends all day working with symbols: "wow! the fundamental nature of intelligence is just like symbols!"

Eric Levitz avatar
⟳ ueaj retweeted
Eric Levitz@EricLevitz·8h

Pretty clear what happened here. TikTok videos make teens anxious and depressed. But AI slop is regenerative for the human soul

I think it's getting harder and harder to claim that the "youth mental health crisis" is related to phones, social media, screens etc. This is a fascinating chart:

Post attachment
DealBook avatar
⟳ ueaj retweeted
DealBook@dealbook·2h

"First, I have to say, comrades, it's great to be here at the largest DSA meeting Lincoln Center has ever seen": @nycmayor has taken the stage at the DealBook Summit

Cassie Pritchard avatar
⟳ ueaj retweeted
Cassie Pritchard@hecubian_devil·4h

So, the problem is that it really isn’t tenable to let anyone download a superhuman hacker to their computer and use it to commit cybercrimes. That’s actually not something that society can tolerate, and the open weights people never had a good answer for this To the extent that solving this problem necessarily privileges a small oligopoly of AI labs, this is a logical foundation for *more* democratic control over this technology, not less. “Yes, because the thing you’re building is too dangerous, we will in fact drive your competitors out of business—but consequently, you have ceased to be a normal firm and we cannot simply apply normal regulatory permissiveness anymore. This is not a free market, by your own design.” That’s a reasonable political premise from which to justify extensive regulation, including “extreme” policies like freezes on development, or nationalization (neither of which is really extreme in the context of a technology that can cause civilizational collapse)

This is where I figured this is all going. Congratulations to OpenAI and especially Anthropic on complete regulatory capture, now that the US open-model community is beheaded, and Chinese open models are on a path to being restricted

ueaj avatar
ueaj@_ueaj·7h

lowk incredibly embarrassing to get beff on this, obviously he does not care about x-risk The Losers' coalition

Ahead of Anthropic’s IPO, a rare coalition across AI safety, open source, e/acc, science, and law has signed the MisAnthropic Statement. Read the statement or sign at: misanthropicstatement.com/

ueaj avatar
ueaj@_ueaj·18h

if I wanted to design a coherent yet expansive magic system I would probably start with a macroscopic quantum mechanics or essentiafied biology, maybe both

Oliver Traldi avatar
⟳ ueaj retweeted
Oliver Traldi@olivertraldi·Nov 17

BREAKING: Philosophy is experiencing a major supply chain outage. Access to things in themselves is limited for all subscribers. Officials claim that information about accounts is still available by inference

ueaj avatar
ueaj@_ueaj·Nov 17

fake fella AI news: ZOMG Claude Astra 7.3 xfast xhigh is .3% better on webslopbench!!! real fella AI news: THIS AI IS FORKLIFT CERTIFIED!!!

Researchers have used reinforcement learning to teach a 5-tonne John Deere skid-steer loader to perform complex acrobatic manoeuvres autonomously. Trained in simulation and transferred to a real machine, the system can balance on two wheels, navigate rough terrain, spin, and control its bucket before landing safely. The research demonstrates the potential of AI-driven automation for heavy construction and industrial machinery. Video Credit: @leggedrobotics / @ETH Paper: wevlv.co/4kltYys #engineering #technology

1:10
taoki avatar
⟳ ueaj retweeted
taoki@justalexoki·Nov 16

here we go with this bullshit again. just because there was christmas last year doesn't mean it's going to happen again

Christmas by the end of 2026

Keegs 🧮🏗️ avatar
⟳ ueaj retweeted
Keegs 🧮🏗️@LittleKeegs0·Nov 16

Heard a "straight" friend of mine say the f-slur so I said "you have to kiss me to say that word" and he did. Woke 2 is here

Byrne Hobart avatar
⟳ ueaj retweeted
Byrne Hobart@ByrneHobart·Aug 6

Pangram is good, but has some well-known failure modes, like: - A time you heard it didn't work - A school essay you wrote years ago that it flagged as AI (you don't have a link or the essay text handy) - A different AI detector got it wrong, and they're all the same, right?

yoya avatar
⟳ ueaj retweeted
yoya@yasuslop·Nov 16

if I milked you I would get goat milk