janhoglund
@janhoglund@mastodon.nu  ·  activity timestamp 5 days ago

…the term hallucinations is subtly misleading. It suggests that the bad behavior is an aberration, a bug, when it’s actually a feature of the probabilistic pattern-matching mechanics of neural networks.
—Karen Hao, Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI
#ai#llms#llm #hallucinations
janhoglund
@janhoglund@mastodon.nu  ·  activity timestamp 6 days ago

”We’re focused on how LLMs hallucinate, but maybe we should pay more attention to the ways we are being encouraged to hallucinate about them. The idea of “artificial general intelligence” or “conscious machines” is more an advertising strategy than a near term engineering reality.”
—Matthew David Segall, A Biophilosophical Dialogue
https://footnotes2plato.substack.com/p/a-biophilosophical-dialogue
#ai#llm#llms
Deborah Pickett
@futzle@old.mermaid.town  ·  activity timestamp 3 weeks ago

I've started using this reading hack to understand articles gushing about Generative AI:

* Substitute "housewife" every time you see a mention of genAI. *

Trying it on this news article from ABC Australia: https://www.abc.net.au/news/2025-06-27/audience-responses-how-generative-ai-impacts-everyday-life/105336182

"I'll often solve problems on a notepad or whiteboard and then have a Housewife generate code and test cases to match the solution I have come up with."

"I created my aunt's eulogy using creative writing with the help of Housewife. The result was a brilliant balance of heartfelt sympathy and sharp, irreverent humour exactly what she would have wanted."

"At home, I use Housewife for all manner of things — find recipes that use specific ingredients (great to help clear out the fridge); assess my Pinterest boards to finally give me a "label" for my decorating style; review different rugs I was considering and tell how/why they fit (or didn't) my style; develop fitness plan; develop menu plans and shopping lists. The list goes on."

"On a more personal level, Housewife has been a quiet support in areas I didn't expect. Whether it's helping me reflect through journalling prompts, composing heartfelt messages for loved ones, or guiding me through mindfulness practices, it's like having a tool that adapts to many facets of my life. Of course, I stay conscious of its limitations, it is not a human, and I always double-check important facts, but as long as I treat it as a collaborator rather than an authority, it adds a lot of value."

#LLM#GenAI
jbz
@jbz@indieweb.social  ·  activity timestamp 3 weeks ago

🫧 The Unsustainable Economics of LLM APIs: Understanding the Coming Price Realignment

｢ The LLM API market has a paradox. While companies invest billions in AI infrastructure, access to these powerful models is priced at levels that seem almost too good to be true. Just like the early days of Uber, it is a subsidized market in a strategic land-grab phase ｣

https://tinyml.substack.com/p/the-unsustainable-economics-of-llm

#ai#llm #aihype
Jeri Dansky
@jeridansky@sfba.social  ·  activity timestamp last month

Sorry for linking to Substack, but this one is so very good:
https://garymarcus.substack.com/p/a-knockout-blow-for-llms

A few excerpts:

Apple has a new paper; it’s pretty devastating to LLMs.

Whenever people ask me why I (contrary to widespread myth) actually like AI, and think that AI (though not GenAI) may ultimately be of great benefit to humanity, I invariably point to the advances in science and technology we might make if we could combine the causal reasoning abilities of our best scientists with the sheer compute power of modern digital computers.

What the Apple paper shows, most fundamentally, regardless of how you define AGI, is that LLMs are no substitute for good well-specified conventional algorithms. (They also can’t play chess as well as conventional algorithms, can’t fold proteins like special-purpose neurosymbolic hybrids, can’t run databases as well as conventional databases, etc.)

#AI #LLM#GenAI#Apple