internetperson

@Aella_Girl true actually. damn

a great test of "would you have been chill with owning slaves, if you'd been born into a slave-owning family" is whether or not you're vegetarian today

@entamechnnel "easy life" working 12 hours a day

@tszzl no researcher would name 4 different products "codex"

the researchers run openai, which is why everything is named so terribly

@sama fine tune on claude

what would you most like to see improve in our next model?

@lukeburgis no it isn't lmao

“Alignment” is not a native category of any serious moral tradition I know. It is a control-system metaphor that frontier AI labs are now trying to convert into a moral anthropology.

@nyutandon i mean, yeah? of course. we don't have AGI yet, this isn't news

Research from Prof Julian Togelius found that despite AI's well-documented victories in chess, Go, and Atari games, humans still learn unfamiliar video games far faster than any AI model. #NYUTandonMade buff.ly/fbVkDrh

@Xxi5olc @arcprize because its easy for humans but AI can't do it, so its clear we don't have AGI yet

@arcprize Can someone please enlighten me about why ARGAGI is really relevant?

GPT-5.5 & Opus 4.7 on ARC-AGI-3 - GPT-5.5: 0.43% - Opus 4.7: 0.18% We found 3 failure modes: - True local effect, false world model - Wrong level of abstraction from training data - Solved the level, didn’t reinforce the reward See our full analysis 🧵

@sama AI is ruined in that case. humans are cursed to work forever

we want to build tools to augment and elevate people, not entities to replace them.

@AnthropicAI did you include my at least 50 conversations where i say some dumb stuff just to see what claude would say in response

How do people seek guidance from Claude? We looked at 1M conversations to understand what questions people ask, how Claude responds, and where it slips into sycophancy. We used what we found to improve how we trained Opus 4.7 and Mythos Preview. anthropic.com/research/claud…

@GregKamradt who's better

ARC-AGI-3 testing is done for gpt-5.5 and opus 4.7 Now we’re in analysis mode going through the logs It’s pretty clear where the failure modes are for each model

@tautologer no

LLMs don't know how they know things but absent episodic memory and explicit reasoning processes, do we know how we know things?

@theCTO @sama it's difficult to do so

hey @sama can we normalize models just saying "i dont know" ? eliminates 99% of hallucinations

@yacineMTB primer if you haven't already seen it. 'The Night Is Short, Walk on Girl' if you like anime

What movie should I watch? No scary movies please

@sama i personally am not going to ve working if possible

"post-AGI, no one is going to work and the economy is going to collapse" "i am switching to polyphasic sleep because GPT-5.5 in codex is so good that i can't afford to be sleeping for such long stretches and miss out on working"

@SenSanders socialists just LOVE to work

Uncontrolled AI poses a severe danger to all of humanity. On Wednesday, I'll be hosting a discussion with leading AI scientists from the US and China about the need for international cooperation against this existential threat. This is an enormously important issue. Join us.

@sama uncancel porn mode already

1. We believe in iterative deployment; although GPT-5.5 is already a smart model, we expect rapid improvements. Iterative deployment is a big part of our safety strategy; we believe the world will be best equipped to win at the team sport of AI resilience this way. 2. We believe in democratization. We want people to be able to use lots of AI; we aim to have the most efficient models, the most efficient inference stack, and the most compute. We want our users to have access to the best technology and for everyone to have equal opportunity. We have been tracking cybersecurity as a preparedness category for a long time, and have built mitigations we believe in that enable us to make capable models broadly available. 3. We love you and we want you to win. We want to be a platform for every company, scientist, entrepreneur, and person. (My whole career has largely been about the magic of startups, and I think we are about to see that magic at hyperscale.)

@p8stie its straight for sure. 4.6 is more gay

The new Claude model is psychologically unbalanced. Reminds me of a sinister homosexual

@dejavucoder i see gpt image is still pissmaxxed

how i had imagined life with coding agents (it's not) (claude, grok, gemini, kimi, chatgpt using gpt image 2)

@yamanakanobody From what I can tell, japanese sleep time is generally shorter

何かどの時間帯でもアメリカ人からコメントが来るけどさ、 アメリカ人は寝ないの？　睡眠時間短い？

@tacowasa2nd its fine. all countries had bad and good people. It's better we see an honest representation of the good and bad parts of foreign cultures, rather than idealized views