Wanna know why Sam Altman and Elon Musk and Peter Thiel all LOVED Eliezer Yudkowsky, right up until they ignored him? It’s because he was USEFUL to their efforts to attract funding and talent to their new companies. If you think futurism is about accuracy, then you’re the sucker at the table.

I like that title+thumbnail much more than the one that the video is showing right now. Definitely need to check this, but I've still not watched the one with Sutton. I'm very glad that someone more knowledgeable than me is saying this, BTW. I don't see why AGI would be a super intelligent thing.

Too many people hear AI and think LLM when no, AI and ML have massive applications beyond LLMs. Just look at the advances in computer vision and audio analysis for instance massive leaps.

Pretty disappointed by this. I did have to answer between answers in which both takes are absurd to me, and I ended up being flagged in a way that doesn't represent me well at all.

I loved the exchange on the other site where someone was sourly dunking on an AI gen implementation of Doom, how John Carmack would hate it John Carmack jumped on the thread and said he thought it was great, then went on about how the game industry needs to use all available tools

Theophite's assumptions are my base case as well: Google and Msft (fine), META (fucked), OpenAI(smoldering crater), Oracle (not fine). Also think companies like Coreweave and Anthropic are also pretty boned. The use case exists, but the business case remains ephemeral because the moats are fake.

this is going to get me yelled at so I will preemptively say "this is not the thread to reply with 'but I just don't like AI and think it's bad'" but I do think that "LLMs are glorified autocorrect" as a meme is actively hindering people's understanding of how this tech works at this point

I’m not anti-AI. I am, however, anti the rapacious tech bros who selfishly hype AI for the purpose of increasing their wealth and power at the expense of our humanity.

It bothers me to no extent that the media plays how Klarna whistles Klarna wants to the headlines to say "AI-enabled Klarna" It would have halved its workforce without AI because it overhired 2x in 2020-21 and never corrected till now But this is how they get front pages

A company that believed it was in the verge of AGI or ASI wouldn’t capitulate to the government because it wouldn’t care about government contracts. They would soon BE the economy and the government would soon be capitulating to them.

I think it needs to be mentioned again and again that the electricity consumption issue is very much real (And especially with future projections) but the water meme is an insane distraction that makes no sense.

the fact that blocking a user hides their reply for everyone, not just the user who blocked them, seems like an extremely toxic pattern for collective sense-making. 1. user makes [xyz claim] 2. another user corrects them 3. blocked, hidden it should be explicitly shown as hidden instead

The Great Software Quality Collapse: How We Normalized Catastrophe techtrenches.substack.com/p/the-great-... Thank you @randomascii.bsky.social for the reference in your latest blog article! 💚

everyone should be able to run one of these on their own. personal moderation really ran by you.

i've always Been This Way

Let’s talk about AI art. theoatmeal.com/comics/ai_art

In general there are two ways to authenticate video in a world with ubiquitous video generation models: a) cryptography: cameras can sign their outputs w/ tamper-resistant modules b) social proof: people vouch for the video (a) is underdeveloped so (b) is about to become very important

*taps sign*

Have to admit there are pro-ai takes on the other site that are as cringe as anti-ai folks on here.

A comment about this article: www.nytimes.com/2025/10/06/o...

Barcelona Open Tech Week is here 🌍 📅 4–9 November 2025 This November, Barcelona will host conversations, workshops and debates, all in the name of open technologies and a better-tech future. canodrom.barcelona/en/opentechw...

Video will become like other forms of testimony. It's kind of shocking how well we compensate for the fact that anyone can just say anything. People lie all the time, but specific testimony is still useful and can be correlated with other evidence.

accidentally caused Nano Banana's system prompt to leak, but in a fun way

anyway, here is 2024 Nobel Prize in Physics winner Geoffrey Hinton discussing what we know about large AI models on 60 Minutes.

it speaks to the level of aggressive, bad-faith and poor quality thinking on here that despite (a) never claiming and (b) outright clarifying several times in the thread, many popular posters on here are intentionally lying and saying that i think that LLMs are “sentient”.

now you of course might know that you’re a real person with real thoughts and real feelings and a whole vibrant interior world. but how can you prove that to a relative stranger on the internet? pictures? generated. video? generated. friends vouching? they’re LLMs too, obviously — i’m not stupid

Not only I agree, but I also happen to have a great example. I've just entered the site to find out to find out that I've been blocked by a person posting this hateful stuff TO ACTUAL HUMAN BEINGS.

Feels good reading a smart person saying this, because it's roughly what I think. Feels bad that Mastodon gets to enjoy the long posts that we don't, and I keep struggling at summarizing stuff with my limited vocabulary and my addition to being super precise.

guy: "hey you guys are being so irrational about large language models without knowing a thing about how they work it's like bigotry/prejudice" people in the replies proceed to demonstrate the truth of his statement by making jokes about 1+1=2(probably) meanwhile

Can't code? I'd say ask a friend, but at the rate AI is improving, maybe you won't even need to. Someone built a functional Appview with replit overnight. bsky.app/profile/doll...

Joshua Rogers, using AI tooling responsibly and professionally, reported 22+ genuine issues in curl that are now being addressed Especially notable because curl had problems with floods of garbage slop AI "security issues" in the past that were nothing of the sort simonwillison.net/2025/Oct/2/c...

🎯 And this is why I don't trust at all the main AI critics when they talk about economics. Yes, tons of AI companies which are mere resellers are going to go down when the bubble pops, yet the big corps can afford to lose a lot on AI, the same way that Google lost a lot on YouTube, for example.

Nice to see another fully open, multimodal LM released! Good license, training code, pretraining data, all here. LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training Slowly, the community is growing. arxiv.org/abs/2509.236...

Most of the "90% of code written by AI" claims come from vendors selling AI tools and lack credibility as a result Armin (creator of Flask, Jinja, Click) is different - when he says 90% of a new significant infrastructure project he's building was AI generated that's worth paying attention to

"We want AI agents that can discover like we can, not which contain what we have discovered. Building in our discoveries only makes it harder to see how the discovering process can be done." - The Bitter Lesson (2019), by Rich Sutton www.incompleteideas.net/IncIdeas/Bit...

In some ways this is like using a Ouija board and seeing what random stuff comes out. But if your Ouija board helps you solve new problems in quantum computing, that’s a pretty good Ouija board!

OK, I really need to try this. I've never done any kind of serious test of any "agentic" tool in the CLI or similar, and the amount of choices is overwhelming. I think I'll bit the bullet an try this hard enough till I can make an opinion on it: www.reddit.com/r/neovim/com...

I know plenty of colleagues who are going to enjoy this a lot. A shame I did not know about this early!

This only happens to you once

And new paper out: Pleias 1.0: the First Family of Language Models Trained on Fully Open Data How we train an open everything model on a new pretraining environment with releasable data (Common Corpus) with an open source framework (Nanotron from HuggingFace). www.sciencedirect.com/science/arti...

it includes model weights + data + training code + paper this is objectively open source

Talking even to academics you often encounter a strong belief that a distinct, new thing called "generative AI" was invented in 2022, while ML tools before that date were hand-crafted and used only trusted sources.

I continue to be amazed at the people who speak confidently about technologies they refuse to use. ChatGPT has had the ability to search the web for over a year now and is, in fact, quite useful for finding links. This post is in response to Google Search AI Overview being shit, lol

many ai dislikers have a model in their head for what ai is and how it works and what limits it cannot exceed and they confidently predict behaviors for AIs all the time which are just wrong, and that you can see are wrong with 5 seconds of effort. but their audience won’t check.

Didn't RT the original post, although it was pretty devastating for Gartner. Am RTing this one, because a) it implicitly confirms the earlier one AND b) it contains a long-sought map precisely detailing the location of Barbra Streisand's house.

Using copyright to violate freedom of speech is very old news for someone like me. But I wasn't expecting it for something so, so small, in a platform like this. I'm very shocked.

Amusing that if you have yet to hear about Claude Code and work in tech: you are either living under a rock, or are someone who keeps up with the latest and greatest with an incredibly expensive Gartner subscription (or thinks they are keeping up, that is)

We'll see how long this lasts. It is quite an important test to see how well BlueSky does in this circumstance, IMO.

We have freedom of speech, and to exercise a criticism, we have the right to show the work under scrutiny. Using copyright to silence this is unacceptable. @bsky.app might be a small player vs Gartner's lawyers, but they should provide a way to contest the absurd copyright claim.

I really don't understand why I can't share this chart to make a commentary about the fact that this chart is flawed information. As Gergely pointed out, it is clearly ignoring quite important players, like Claude Code and Codex. There are many others, but the lack of this two is suspicious.