llm

This #genai #llm #tokenCost is getting silly. Can't wait for the bubble to burst.

Researchers just mathematically proved that AI can't recursively self-improve its way to superintelligence.

Not "we think it's unlikely." Not "it seems hard." Formally proved.

The model doesn't climb toward AGI — it slowly forgets what reality looks like. They call it model collapse. The math calls it inevitable.
I wrote about it 👇

https://smsk.dev/2026/04/26/ai-cannot-self-improve-and-math-behind-proves-it/

#AI #MachineLearning #LLM #Research

Interesting.

OpenAI Misses Key Revenue, User Targets in High-Stakes Sprint Toward IPO

The company’s CFO and board have questioned the wisdom of massive data-center spending in the face of slowing growth

From https://www.wsj.com/tech/ai/openai-misses-key-revenue-user-targets-in-high-stakes-sprint-toward-ipo-94a95273

Several interesting details in this article, including that both Google and Anthropic are eating a bit of OpenAI's lunch. DeepSeek's latest model would be poised to eat all their lunches if it weren't for the US government running cover for US firms. This does not feel like a stable situation to me.

#AI #GenAI #GenerativeAI #LLM #OpenAI #ChatGPT #economy #economics #Anthropic #Codex #Claude

I am impressed at how they shoved LLM bullshit into a fucking website that just shows links. I said this in a reply, but just saying. Your own website will never do this to you, I promise. https://linktr.ee/help/en/articles/9269822-ai-features-on-linktree #LLM #AI #LinkTree

"Writing is thinking" 👈 Excellent!

📝 https://doi.org/10.1038/s44222-025-00323-4

I have always seen #writing as more than mere expression. It constitutes active #cognition, integrating #memory, #reasoning, and meaning into a coherent structure. Delegating this process to eg #LLM risks reinforcing the opposite: passive consumption instead of structured thought.

A large international study coordinated by the #EBU and led by the #BBC found that AI assistants misrepresent news content 45% of the time across different languages and platforms, with #Gemini performing the worst.

[…] Key findings: 

• 45% of all AI answers had at least one significant issue.
• 31% of responses showed serious sourcing problems – missing, misleading, or incorrect attributions.
• 20% contained major accuracy issues, including hallucinated details and outdated information.
• Gemini performed worst with significant issues in 76% of responses, more than double the other assistants, largely due to its poor sourcing performance.
• Comparison between the BBC’s results earlier this year and this study show some improvements but still high levels of errors.

https://www.bbc.co.uk/mediacentre/2025/new-ebu-research-ai-assistants-news-content

#aihype #llm #openai #perplexity #chatgpt

Watching @blender shoot itself on both feet, and sinking into the slop mire is both sad and amazing, at the same time.

Much in the proverbial train wreck manner.

Anthropic "donates". Anthropic "requests" seats on the board. Anthropic takes over. Blender is used by Anthropic for nefarious purposes. Blender becomes an empty shell, vaguely remembered by those who used it BEFORE the slop.

#LLM

Somehow I have a very strong feeling that within a little time..

Companies are gonna beg for developers to fix their #AI mistakes they making right now

Little tip of advice, don’t

Let them dwell in their AI misery, it’s what they deserve

#LLM

I thought this was a particularly good analysis of the problem of using LLMs for science. It explores the purpose of science, the perverse incentives that drive people to use LLMs, and the impact this has on skill building and training future scientists.

My lab group has been struggling with this topic lately, without much consensus. This blog post captures a lot of our thinking, and very clearly made some good points that we appreciated. It mostly just describes the mess we're in without offering much useful advice, but just laying out the problems do nicely is helpful. That said, I do worry the author may be underestimating the impact these tools might have on experienced researchers.

https://ergosphere.blog/posts/the-machines-are-fine/
#academicchatter #llm

I have an opinion about AI, which is that the LLM scam has yet to emerge. I know a lot of people on the fediverse are just automatically against LLMs and with what seems like a good set of reasons, but really they’re lacking in detail and just going along with a tide of opinion.

For most fediverse ‘LLM haters’ (or AI haters, but not all AI is an LLM) they’ll point to outrageous use of water for cooling, outrageous use of electricity, huge amounts of processing needed, insane amounts of GPU requirements, and so on. This is not even to mention the dodgy position on original effort and copyright and ownership of matter used in training. All seems bad. And it would be if this were to be believed.

But for a moment, let’s just not believe it.

The processing requirements - yes, you need high power GPUs for training. Why are we continually training LLMs? Most of the initial training was done long ago. Much of the training now is done by ‘inheriting’ the training from the initial models and refining it a lot, adding amounts of further training on top. Are we still training models as much as we used to be? More? Not as much? As much but more efficiently? Do we have any actual reporting on this as we stand today?

The amount of GPU needed (and consequently electricity) seems to me the thin end of a wedge-shaped iceberg of rabbit mines. Yes you ‘need’ a GPU to plough through a lot of numbers. Quite why we still need a ‘graphics processing unit’ to perform vector calculation is a quaint mystery of terminology – an early movement toward general purpose computing using the parallelism of GPUs used for shader calculation shifted the GPU away from purely graphics processing but didn’t really bring forth a recognised parallel computer platform (in the way we nearly had with the transputer in the 80s).

Today the tensor calculations used in AI and LLMs are pumped through GPUs. Perhaps we’re waiting for a massively parallel computer platform using optical computing. In the meantime, scaled-up parallel vector and tensor processors will have to do. Why though? Is that where the cartel money is, in keeping the GPU trade exactly as it is, only more of it? Surely something can come in from the side unexpectedly to eat this opportunity?

More than this though, I strongly guess or suspect that the amount of investment being sucked into the AI industry is completely describable as organised crime. That much money is extorted on the basis that huge data centres are required and that they’ll need to be filled with an arrangement of running GPUs and this is where the investment money will end up. I reckon that is mostly bullshit.

The data-centre argument is because the current ‘LLM as a Service’ model needs to attend to many millions of users all at once, and their prompt processing needs to be handled with minimum delay. We get that, yes. So it needs to be better than a standard web server with a complex web site. Yes. But wait, does it really? Really?

I don’t think the future is in AI as a Service so much as we’re being convinced it is, I think the future is more in local powerful models, not one big centralised provision. This is perfectly understandable – except if you’re behind running a big centralised provision, in which case your propaganda will laugh at local models and of course promote the single big ‘as a service’ provision.

I also doubt that a lot of the training is actually what they say it is. A lot of new models are trained on the results of old models. Not copied, but using an existing model, or more than one, to train ‘from’. Most of the energy use of LLMs I suspect has already happened, and there isn’t the need for an awful lot more. Except if you’re already structured for investment money funnelling in because training.

I also doubt that a lot of the results given in a ‘as a service’ model live over the centralised server provision is artificial – I am guessing that some of it is human-faked. And that’s where a lot of the investment money goes to – into paying humans to supplant deficiencies in the LLM scheme when run live for the entire population of the world that wants to use an online LLM service.

A local LLM can certainly do all that is required, but when the whole world wants to use a big centralised one, I’m guessing that having human assistance behind the scenes becomes necessary to ensure it all flows in time. Do we have any reporting on this front, I wonder?

I think the demands of LLMs – the training time and energy, the electricity to run it, the RAM, the GPU, etc, are real but the amount is inflated and when applied to centralised ‘as a service’ business structured projects, has been allowed to be outright deceptive at the least, and at most, almost mafia-like in terms of extortion, like a protection racket, designed to keep the already outmoded ‘centralised LLM as a Service’ con going.

Anyway, this is purely my guess. I’ve not looked into this at all.
#AI #LLM

Software is inert without a human user, even though modern-day "agentic" apps are doing their best to squeeze human agency out of the loop. The focus on the tech and not how it is used has intensified an old phenomenon: interpassivity.

The vision tech has been pushing (software that "uses itself") has created a synthetic replacement for human agency. It is harder and harder to care, and that is by design.

https://productpicnic.beehiiv.com/p/the-newest-term-in-the-ai-lexicon-is-interpassivity

#AI #LLM #tech #software #UX #UXdesign

Had a blind person tell me today, offline, that they actually hope LLMs eliminate the need to go to websites because that would mean he would never have to fight an inaccessible website again, and I'm just sad forever now. It was somewhat similar to https://www.uxtigers.com/post/accessibility-generative-ui #AI #LLM #WebDev #Accessibility

I just read this article about chess-like vs poker-like problems, and how poker-like problems are bad candidates for solution with LLM type models.

I'm looking for more reading like this about the underlying architecture, behavior, and future directions of the models and the cutting edge research around them. Does anyone have links they found valuable?

https://www.latent.space/p/adversarial-reasoning

#noai #aicritic #ai #llm

[2606.03811] AI Agents Enable Adaptive Computer Worms
https://arxiv.org/abs/2606.03811

Forget all previous threats. Forget Mythos.

Given the current state of the infrastructure and all the bugs in firewalls, perimeter security doesn't protect you seriously.

If you can't turn off the computer because it controls your critical infrastructure you have a hard problem now.

#kritis #security #ai #llm #worm

In other words Ladybird is no longer a free and open source software project. It is, in an important sense, closed now.

"Proprietary" means having owner(s). If the maintainers are the only ones allowed to introduce changes, they are establishing an ownership relationship over the codebase. I.e., it's a proprietary project, and that is the sense in which it's closed.

This is the destiny of most projects that go AI, in my opinion.

https://exquisite.social/users/thomholwerda/statuses/116697078838020135
https://ladybird.org/posts/changing-how-we-develop-ladybird/

#AI #GenAI #GenerativeAI #LLM #web #WebBrowser #browser #tech #Ladybird #FOSS #OSS #OpenSource #ClosedSource #ProprietarySoftware

Good stuff from @mitsuhiko about clanker-generated bug reports, with very specific gripes and a strong, deep conclusion:

https://lucumr.pocoo.org/2026/5/24/pi-oss/

#GenAI #llm

The Morse Code Hack That Made an AI Agent Spend $200,000

Dave explains the Grok/Bankrbot exploit that caused an AI Agent to spend almost $200K in tokens!

https://www.youtube.com/watch?v=UQ4pSVS_mN0

#AI #LLM #Grok

Alright I better announce this actually. At 8UTC Sunday 10th May ("tomorrow, Sunday morning in Europe") I am speaking to

@bagder of #curl https://curl.se/ https://en.wikipedia.org/wiki/Curl_(software)

about becoming targetted by trillions of dollars of #AI companies #cybersecurity scanning, especially after he rejected their ai-content merge requests. And having to close bug bounties due to #llm spam.

...And what it means for #indie #programming today. #commonLisp #ecl 's 2010 example is curl, and and and.

Meta Smart Glasses Empower Blind People to Participate in Corporate Surveillance https://thesqueakywheel.org/meta-smart-glasses-empower-blind-people-to-participate-equally-in-corporate-surveillance/ #AI #LLM #Blind #Disability

I was only made aware of this (frankly awesome) case of LLM poisoning today: https://www.nature.com/articles/d41586-026-01100-y. A researcher made up a disease and published two evidently fake preprints about it (including sentences such as “this entire paper is made up” and “Fifty made-up individuals aged between 20 and 50 years were recruited for the exposure group”), which were almost immediately picked up by LLMs and documented in their output. Worse, actual – supposedly serious – medical papers also started citing the preprints, demonstrating that academics relying on LLMs to do their work is a genuine problem! Not that I had my doubts but, if anyone did, this seems like the perfect demonstration of the problem. Article immediately added to the syllabus of the class I am co-teaching with Iris Ferrazzo on LLMs for Romance Studies/Humanities!

#LLM #GenAI #academia #research #ResearchIntegrity #humanities

I think the acceptance of AI/LLM as valid programming
tools is partly a symptom of programming becoming
de-skilled and hype driven over like the last 10-15 years.

First "everybody gets to be a programmer" and secondly
this has set the stage for "this pattern matching machine
gets to be a programmer".

#LLM #genAI #agentic

I type those tags and i want to type "crap"
after each one:
#LLMcrap #genAIcrap #agenticCrap

Another is reason is that people that are past their primes
as programmers but have lots of clout and $$$ want to
stay relevant to the hype machine.

(I don't think programmers necesarilly go "past their prime",
I just mean that specifically for the several high profile
programmers that are shilling for genAI.)

I do not believe that if the major AI companies actually had the technology they suggest they have that they would sell it to the public for even $2,000/month.

#AI #GenAI #GenerativeAI #LLM

Gradient ascent is superior to generate-and-test in any domain in which you have even a heuristic gradient that is sometimes wrong.

Why people working on software where something serious is at stake would throw out known gradient to use a code generator + testing is beyond my capacity to understand.

https://1password.social/@1password/116580082041363054

#AI #GenAI #GenerativeAI #LLM #VibeCoding #Software #SoftwareDevelopment #tech #dev #security #InfoSec #PasswordManagers

Watching techies learn that their favorite artist isn't in love with LLMs will never be boring to me, online or offline. The best part? The techs act all shocked their favorite artist would hate something that plagiarizes from them by the minute. Of course, the rebuke always is, well, artists don't understand how tech works anyway so they shouldn't be listened to. But still, the initial reaction is priceless, every time. I was at a dinner and the general feeling was just utter shock that the tech Enthusiasts artists, some fantasy author, would be against his loved generator. When I tried to explain why, he instantly assumed I didn't know how tech worked because I don't breathe in Rust or I don't Exhale in JavaScript. #LLM #AI

Y’all, we need to talk about upcoming #IPOs, and the insane rule changes that #nasdaq has just announced.

Nasdaq rewrote its #index inclusion rules to accommodate #SpaceX’s mega-IPO, implementing a "Fast Entry" provision that allows the company to join the Nasdaq-100 index just 15 trading days after its initial public offering, down from the standard three-month seasoning period.

They also eliminated the minimum float requirement of 20% available public shares and instead stocks with less than 20% of shares publicly traded, Nasdaq applies a 3x multiplier to the free-float for index weighting purposes, artificially inflating low-float giants like SpaceX in passive funds.

Ok, but in English? #SpaceX, #OpenAI and #Anthropic have just figured out a scam to force every passive #IRA, #401k, and index fund to buy their stock before pricing evaluation.

They’ve figured out how to steal your #retirement.

#AI #LLM #Scam #guillotines #YouWillOwnNothing

https://www.businessinsider.com/spacex-ipo-s1-spcx-stock-nasdaq-qqq-elon-musk-2026-5