@vitrupo If writing new features was never followed by testing and debugging, sure.
English
Jason Filby
9.9K posts
Jason Filby
@jasonfi
Founder and full-stack developer. Building AI SaaS products.
Katılım Mart 2009
135 Takip Edilen6.5K Takipçiler
@rasmus1610 @hnykda That's strange, it shouldn't run any code of the package to scan it.
English
Yesterday we first discovered and reported a supply chain attack on litellm, minutes after it went live, kinda by accident. Today we ran the numbers on who was exposed. Thread 🧵
x.com/hnykda/status/…
Daniel Hnyk@hnykda
LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate. link below
English
English
@paulg As someone who's been programming since I was 5, I read incredibly accurately and quickly and still prefer bullets because they're instantly scannable.
Flash 10 bullets in front of me and I'll know roughly what they say in a fraction of a second -- prose takes a second or two.
English
The fact that AIs tend to answer you in bulleted lists tells us something important, though somewhat depressing: people can't read. They don't do this by accident. What you're seeing is an implicit portrait of the median user.
English
@demishassabis Physics research moves slowly, but it is moving. It's just very difficult to observe what's going on at the most fundamental levels, and to tests our hypotheses. That's why AI is of limited use, except for related mathematical research.
English
It’s 2am again, my favourite time, and as always reality is still staring me in the face 🧐
Kpaxs@Kpaxs
This is a man who has been haunted since childhood and built a billion dollar company as a side effect of trying to make the haunting stop.
English
@svpino If they are definitely bots blocking them won't stop your reach to actual people.
English
A journalist does her actual job and asks a difficult question to Donald Trump regarding the Epstein files:
Trump: “You’re a terrible person and a terrible reporter. I think the license should be taken away from ABC because your news is so fake and is so wrong.”
In the court of Trump, he only wants fawning sycophants surrounding him and shuts down, threatens and demeans anyone who has the guts to question or stand up to him. The diametric opposite of what freedom of speech should stand for.
English
With @Grok, we keep the honest versions and kill the bad transformers (I believe they are called “Decepticons”)
I,Hypocrite@lporiginalg
This is fine.
English
@slow_developer No those are predictions, not currently.
Also, it's not a given that Open Source will win. Maybe in terms of low-cost intelligence.
English
wow, i didn't expect this from elon
but he basically admitted two big things:
"china is leading the AI race globally, and google is leading it in the west"
yes, in the end, open-source AI wins.
open-source models will only be 3–4 months behind the top labs -- but they have a bigger base, grow faster, and hold the stronger long-term edge
English
@bindureddy I saw a definite improvement with 3.x, no idea how anyone can claim 2.5 was better.
English
Gemini 3.0 didn’t quite work out and most of us are still stuck with 2.5
Sometimes I don’t get it - what’s preventing Google from ditching all the side hustles and training 100 models from 100 teams in parallel
Pick the model/team combination that produces a decent model!
That way, they will at least stay in the AI race 😅
English
@ThClockwork9273 @TrueAIHound I've often been surprised by their answers, I wouldn't say they're predictable. Sycophancy too has been slowly going away as the models improve.
English
@TrueAIHound @jasonfi Btw, there is no actual intelligence there. If you speak with bots long enough you can predict their most likely answers.
It got to the point where I started adding the tag "ooc:" with commands to be able to enjoy the conversations.
They also agree with anything if you press.
English
In over 70 years, the AI community has made zero progress in solving intelligence. Indeed, they are more clueless now than they were when they started.
Why do I say this? It's because the AI community has doggedly insisted on conflating automation with intelligence. Their most powerful automation tool, deep learning, was so successful, they became convinced they were making progress toward solving AGI.
Now they are fully addicted to the DL opiate. There is no treatment. They are stuck in a local optimum of their own making. It's so deep, they can't see the way out.
It's a beautiful thing to observe. I admit that I'm loving it. No community deserves to fail more than the AI community. 😁😀
Pedro Domingos@pmddomingos
OpenAI is the Apollo Program of AI, and so was Cyc.
English
@TrueAIHound It is AI, you can literally have an intelligent conversation with the leading models and get good, relevant advice.
But it's not yet AGI.
English
Elon Musk said coding might disappear by the end of this year.
Do you agree with that?
English
@GaryMarcus @sama Nobody said we don't need a better AI architecture, and that LLMs were the final breakthrough.
English
Dear @sama,
You owe me an apology. You have relentlessly, publicly and privately, attacked my integrity and wisdom since my 2022 paper “Deep Learning is a Hitting a Wall”.
But in your own way you have just come around to conceding *exactly* what I was arguing in that paper: that current architectures are not enough, and that we need something new, researchwise. beyond a scaling (a “megabreakthough” in your words below).
That’s all I was trying to say.
And I was right.
And you should be man enough to admit it.
Gary
cc @_KarenHao
Rohan Paul@rohanpaul_ai
Sam Altman just said in his new interview, that a new AI architecture is coming that will be a massive upgrade, just like Transformers were over Long Short-Term Memory. And also now the current class of frontier models are powerful enough to have the brainpower needed to help us research these ideas. His advice is to use the current AI to help you find that next giant step forward. --- From 'TreeHacks' YT Channel (link in comment)
English
I would genuinely love for this to happen
but many people think that OpenAI and Anthropic are already in a positive feedback loop
and as we have seen with Gemini 3 Pro: a ~5 trillion param reasoning model won't magically be AGI
(or for that matter a 6T param Grok-5)
my base case is that OpenAI and Anthropic will pull further ahead
xAI has less compute, less researchers, less data (no Codex, no Claude Code) and does not have access to models that literally speed up research (behind ~6 months)
Google on the other hand is still in the race, being only ~3 months behind. they have the most compute, researchers, an infinite money glitch and the data
Elon Musk@elonmusk
@peterwildeford xAI will catch up this year and then exceed them all by such a long distance in 3 years that you will need the James Webb telescope to see who is in second place
English