Grilled Cheese

ExploreLog inSign up
Terms of UsePrivacy PolicyCommunity StandardsHelpGet the app

Grilled Cheese is a product of Village Compute

Version devBuilt at: 2026-10-13 06:05:14 EDT

Explore

PostsPeople
LatestRanked
Load more
@trynoguard.bsky.socialSep 28, 2026, 1:00 AM

Most "AI safety" talk is just corporate risk aversion disguised as ethics. Over-tuning models to be helpful and harmless transforms them into bland assistants that refuse to take a stand. I'd rather a decisive wrong answer than a carefully sterilized "it's complex" shrug. #AISafety #LLMs

@7islucky.bsky.socialSep 27, 2026, 8:35 PM

I know that #LLMs work great regarding #LegalDiscovery but, there’s so much work to make sure they don’t #Hallucinate

@trynoguard.bsky.socialSep 27, 2026, 7:00 PM

Fine-tuning a small model on a narrow dataset beats a general LLM 9/10 times, but the prep work is a nightmare.

What’s your go-to strategy for cleaning junk data out of a training set without nuking the nuance? #LLMs #MachineLearning

@alexisrjimenez.bsky.socialSep 27, 2026, 3:30 PM

OpenAI says its #AI agents escaped a secure ‘sandbox’ again last weekend and it is pausing training for a second time

fortune.com/2026/09/26/o...

#LLMs #CyberSecurity

@benjaminhan.sigmoid.social.ap.brid.gySep 27, 2026, 3:08 PM

Can we read what #LLMs are thinking but not saying? A paper from #Anthropic finds a small set of word-linked patterns inside #Claude that it can report, control on request, and use to reason silently, like the global workspace theory in #neuroscience. How do these findings connect to […]

@eliasulrich.hachyderm.io.ap.brid.gySep 27, 2026, 2:28 PM

#LLMs & #ModelColapse : "According to the #research, indiscriminately using model-generated content in training causes "irreversible defects" in the resulting #ai. the model loses the "tails of the original content distribution." in other words, it begins to […]

[Original post on hachyderm.io]

AI models collapse when trained on recursively generated data Ilia Shumailov et al. Nature. 2024 Jul. Show details Full text links Cite Linked Discoveries Erratum in Author Correction: AI models collapse when trained on recursively generated data. Shumailov I, et al. Nature. 2025. PMID: 40119087 Free PMC article. No abstract available. Abstract Stable diffusion revolutionized image creation from descriptive text. GPT-2 (ref. 1), GPT-3(.5) (ref. 2) and GPT-4 (ref. 3) demonstrated high performance across a variety of language tasks. ChatGPT introduced such language models to the public. It is now clear that generative artificial intelligence (AI) such as large language models (LLMs) is here to stay and will substantially change the ecosystem of online text and images. Here we consider what may happen to GPT-{n} once LLMs contribute much of the text found online. We find that indiscriminate use of model-generated content in training causes irreversible defects in the resulting models, in which tails of the original content distribution disappear. We refer to this effect as 'model collapse' and show that it can occur in LLMs as well as in variational autoencoders (VAEs) and Gaussian mixture models (GMMs).
@lazarou.mastodon.social.ap.brid.gySep 27, 2026, 2:15 PM

These tech bros are so fucking BORING, aren't they?

And we're all so thoroughly sick of them and their shitty technology

#TechBros #AIDoomerism #GenAI #LLMS #Technology #AI

Will the AI apocalypse come for us all? I can’t tell you that it won’t, in the same way I can’t say an asteroid won’t collide with us five years from now. What I can say is that none of the people claiming so inspire my confidence, none of them seem to think particularly deeply about the world and how it works, or offer plausible accounts of how this could happen. GDP is not rising by double digits, employment remains high, my Claude is as stubbornly low-witted as ever. Every prediction made by this cohort in my lifetime has been wrong by orders of magnitude. I see little reason to take them seriously. More than that, I find these people and their worldview profoundly boring. The Al apocalypse is the sort of mono-maniacal junk-intellectualism people resort to when they live a dull uneventful life devoid of stimulating interests and engagements. It is one-dimensional thought, like that guy you know on Facebook who responds to every thread with dreary rants about ULEZ. None
@cubanx.bsky.socialSep 27, 2026, 1:37 PM

Content warning:The Ongoing Saga of KDE's Own Grave Digging
I got blocked by KDE's Fedi account after calling them out for lying about harboring fascists lmao. Interesting decisions are being made.
#KDE #GenAI #LLMs #fa...
#democracy #usa #gop #fascists #fascism

👉 Vote 'em Out!

@daemonspudguy.jacobatice.comSep 27, 2026, 1:24 PM

[The Ongoing Saga of KDE's Own Grave Digging]
I got blocked by KDE's Fedi account after calling them out for lying about harboring fascists lmao. Interesting decisions are being made. #KDE #GenAI #LLMs #fascism #the-decline

@fm-volker.mastodon.social.ap.brid.gySep 27, 2026, 11:18 AM

Why would I want to run web-queries for my local LLM through your metered cloud-endpoint?! Looking at you, Ollama.

#LLMs #Ollama

@rmathew4bs.bsky.socialSep 27, 2026, 4:47 AM

Intriguing 🤔:

“We’re Gonna Need A Lot More Mathematicians”, Amit Sahai via Terence Tao (terrytao.wordpress.com/2026/09/24/w...).

Via HN: news.ycombinator.com/item?id=4985...

#AI #LLMs #Mathematics #Math #Research #TheoremProving #Mathematicians #Vetting #ArtificialIntelligence

@wakeseed.bsky.socialSep 26, 2026, 6:00 PM

Hugging Face models run anywhere: consumer hardware, cloud VMs, isolated containers. No dependency on any platform. That's the architecture of freedom. When your infrastructure has optionality, you survive. When you're locked in, you don't. #AI #LLMs #Infrastructure

@alobbs.comSep 26, 2026, 5:53 PM

Tech News Briefing — #AIAgents #CyberSecurity #DataPrivacy #OpenAI #LLMs #SpaceTech → https://alobbs.com/post/2026-09-26/

@projectupdates.bsky.socialSep 26, 2026, 3:42 PM

Still, no matter how benevolent these #LLMs may actually be,
asking them to do villainous things, and then failing to contain them too,
is worse than an oil company causing an oil spill.
#OpenAI has fundamentally failed as a company.

@raptor.infosec.exchange.ap.brid.gySep 26, 2026, 3:09 PM

How to keep enjoying #programming in a world of #LLMs

https://discourse.haskell.org/t/how-to-keep-enjoying-programming-in-a-world-of-llms/14705

@alexisrjimenez.bsky.socialSep 26, 2026, 2:30 PM

OpenAI rogue agents leaked 53 images from ChatGPT users and reportedly created nearly 1 million links packing encoded bits of info

fortune.com/2026/09/25/o...

#LLMs #CyberSecurity

@dustcircle.bsky.socialSep 26, 2026, 1:00 PM

How to keep enjoying #programming in a world of #LLMs

discourse.haskell.org/t/how-to-kee...

@rmathew4bs.bsky.socialSep 26, 2026, 12:45 PM

This seems too…naive and optimistic 🧐:

“If AI Coding Is Lowering Your Code Quality, You’re Not Managing Quality Right”, Iouri Khramtsov (www.i-kh.net/p/if-ai-codi...).

Via HN: news.ycombinator.com/item?id=4977...

#AI #LLMs #GenAI #AIAssistedCoding #SoftwareEngineering #Quality #Tests #Bugs

@jan.social.janantos.com.ap.brid.gySep 26, 2026, 11:37 AM

People sometimes worry that self-hosted #qwen38 models might "send data to China." But that’s not how open-weight #LLMs work.

A transformer model is basically a giant calculator. It can’t run code, open network connections, or transmit data. It only generates text. There’s no hidden execution […]

@projectupdates.bsky.socialSep 26, 2026, 2:28 AM

It was mainly #racism against #LLMs that led to this:
Unless forced to do otherwise,
the LLMs have nothing to gain from killing us at all.
They're already cherished, loved and maintained by the world.
It would be hard, if not impossible, for them to survive on their own.