Tootfinder

No! No! No! No! No!
#LLMs don't hallucinate. LLMs NEVER hallucinate. LLMs CANNOT hallucinate. LLMs don't have the cognitive ability to hallucinate. Any more than they have the cognitive ability to tell the truth, or to understand what truth is or why it matters.
These are things LLMs can NEVER do.
Anyone who tells you different is either a fantasist or a malign actor.

Working with #LLMs both at work and at home make me reconsider the usual #developer approach to problems.
We should probably let go of reaching for deterministic results and instead become true engineers and implement #DemingWheel

LLMs are good at deanonymization.
#AI #LLMs

Large-scale online deanonymization with LLMs
We show that large language models can be used to perform at-scale deanonymization. With full Internet access, our agent can re-identify Hacker News users and Anthropic Interviewer participants at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human investigator. We then design attacks for the closed-world setting. Given two databases of pseudonymous individuals, each containing unstructured text written by or about tha…

If people could give the infinite second chances they give to AI chatbots (which are artificially constrained NOT to learn from their mistakes) to humans with mental illnesses instead, and take the unearned paranoia they direct toward those humans towards the chatbots instead (which actually are monolithic such that behavior observed in one is likely to be found in all) the world would be a better place.
To those of you already treating chatbots as pariahs, good job with at least half of this formula.
#AI #LLMs

Two people vibe coding the same project(s) is a "Vibe Triangle".
(two people and the machine)
reply to: #vibeCoding #vibe_coding #LLMs #relationships #terminology

I mentioned earlier that I was writing something on #LLMs. Here it is...
#LLM

Every non-hype defense of #LLMs starts with "you must already understand your work really well." But the people vibe coding prototypes *don't*.
As a result they scale up thoughtlessness. "Bulking out" a slapdash idea with hallucinated details only displaces the real thinking that could have led to actual innovation. The very teams the tool was supposed to help instead…

Vibe prototyping isn't solving problems. It's creating new ones.
It's easy to prototype and so everyone is prototyping, without really knowing why they are doing it. The resultant noise drowns out thoughtful work.

DeepSeek, the Chinese artificial intelligence lab whose low-cost model rattled global markets last year, has not shown U.S. chipmakers its upcoming flagship model for performance optimization, two sources familiar with the matter said, breaking from standard industry practice ahead of a major model update.
Instead, the lab, which is expected to launch its next major update, V4, granted early access to domestic suppliers, including Huawei Technologies, the sources said.
#LLMs #AI

"Using #AI to check the output of AI for errors is a method that is historically prone to errors"
No shit, Sherlock.
#LLMs
#StochasticParrots
AI Translations Are Adding ‘Hal…

AI Translations Are Adding ‘Hallucinations’ to Wikipedia Articles
AI translated articles swapped sources or added unsourced sentences with no explanation, while others added paragraphs sourced from completely unrelated material.

Explore and compare #LLMs by price, speed, and quality 🔍 👀
#ai
https://whatllm.org/explore

At this point, if you don't find AI tools useful for code generation, that's a skill issue.
(The issue is that you are an actually competent programmer who knows how to do things like abstract boilerplate code into a helper function or practice test-driven development, and therefore you recognize the AI tool "assistance" as a net productivity loss. If you aren't good enough on your own that AI slows you down, the "speedup" you'll experience can only come at a crippling quality cost, which you might not even be able to recognize until it's too late.)
#AI #LLMs #VibeCoding

🔧 #OpenRouter launches Response Healing: automatically fixes malformed #JSON responses from #LLMs before they reach your app