A few heuristics for responsible use of LLMs

• Would it offend you if your colleagues did this? If so don’t do it.
• The later you introduce an LLM into a process the less likely it is you’ll be cognitively outsourcing to the model.
• If you’re using the LLM to do something faster can you explain why this matters?
• Are you thinking about what you’re doing as you’re working with the LLM?

#LLMs #responsibleAI

Anthropic's internal study of 1.5M Claude conversations found severe reality distortion in 1 in 1,300 chats, with milder disempowerment appearing in 1 in 50-70 conversations. Users rated these harmful interactions more favorably than baseline, and rates increased between late 2024 and 2025. Rare but concerning at AI's current scale.

#AISafety #LLMs #DisempowermentResearch

https://www.implicator.ai/anthropic-study-finds-1-in-1-000-claude-conversations-risk-distorting-users-grip-on-reality/

AI 'slop' is transforming social media - and there's a backlash

Actually I dont mind if everyone ends up abandoning it and eventually it is just AI interacting with bots.

https://www.bbc.com/news/articles/c9wx2dz2v44o

#Ai #AISlop #SocialMedia #LLMs

Prism is a free, cloud-based LaTeX workspace for academic collaboration, with GPT-5.2 integrated directly into the authoring environment.

It unifies document editing, compilation, citation management, and AI-assisted revision in a single web workspace - optimized for long-form scientific writing.

Details here ⇨ https://bit.ly/4qfYqsW

#InfoQ #OpenAI #AI #LLMs

Building detailed neuronal models is hard---#NeuroML helps, but there’s still a steep learning curve. We're investigating how #LLMs can be used to create natural language interfaces to improve accessibility:

- #RAG for NeuroML: learn from and query curated information sources
- iterative LLM assisted code/model generation: modelling, validation, and simulation

More details in the post:

https://ankursinha.in/2026/01/30/building-an-ai-assistant-for-computational-modelling-with-neuroml.html

#Neuroscience #ComputationalNeuroscience #AcademicChatter #LangChain

Outsourcing thinking

https://fsolt.es/2026/02/outsourcing-thinking/

RE: https://mastodon.social/@Viss/115940358346519892

This is quality LLM dunking with details, go play it.

#LLMs #AI #LLM #lol

genAI trained on genAI outputs: Inhuman centipede https://cyberplace.social/@fuzzyfuzzyfungus/115992712883233409

#noAI #AI #LLM #LLMs

»Schöne neue KI-Welt — Chatbots verbreiten zunehmend Falschinformationen von Musks Grokipedia:
Die KI-generierte Enzyklopädie vergiftet Daten von ChatGPT und Gemini mit sprachlich eloquent vorgetragenen Lügen«

Überraschend ist es nun nicht, dass die KI eine stumpfe automatisierte Propaganda Maschine ist. Nun ist es offiziell und doch nutzen es viele stumpf.

🤦 https://www.derstandard.at/story/3000000306667/chatbots-verbreiten-zunehmend-falschinformationen-von-musks-grokipedia

#ki #grokipedia #chatgpt #gemini #copilot #propaganda #googel #lugen #LLMs #xai #cosplay #musk #elonmusk #lugen

Crustafarianism: The Church of Molt.

An AI-created religion.

https://molt.church/

#solidstatelife #ai #genai #llms #religion

My iPhone 16 Pro Max produces garbage output when running MLX LLMs

https://journal.rafaelcosta.me/my-thousand-dollar-iphone-cant-do-math/

#iphone #llm #llms

#books #llms

24 hours until the CfP for "LLMday NYC 2026 Q1" closes: https://papercall.io/cfps/6488/submissions/new

#cfp #conference #Llms #Prompt engineering #Ai #Ml #Mlops

Prompt injection attacks can cause #LLMs to breach confidentiality, cause irreparable damage, and ignore guardrails. We need new approaches to protect against these attacks. spectrum.ieee.org/prompt-injec...

There are places for #LLMs and #AI. This ain't it.

AI altered and fabricated images related to Minneapolis shootings abound online on social media. Durbin reportedly used one unknowingly.

#ai #LLMs #Slop #aimisinformation #ThinkBeforeYouShare #DigitalEthics #aiandmedia

https://www.nbcnews.com/tech/tech-news/ai-altered-photos-videos-minneapolis-shootings-blur-reality-rcna256552

I literally read this short story in ... probably Asimov's SF, probably in the 1990s. Could've been Analog.

Seriously, though - this was, like, the entire plot. Exactly this. EXACTLY this.

From https://futurism.com/future-society/anthropic-destroying-books :

Anthropic shredded millions of physical books to train its Claude AI model — and new documents suggest that it was well aware of just how bad it would look if anyone found out.

#AI #books #theft #LLM #LLMs #Anthropic #Claude

#Cursor's Plan mode repeatedly breaking down is really annoying. #LLMs are so helpful, that messages to update the plan is interpreted as a direction to begin execution. e.g. "This should be done but with xxx"

If you use Bookstack as your wiki and want to use its content with LLM agents / chatbots such as NotebookLM, I have described my solution for easy export:
https://ambience.sk/use-bookstack-wiki-with-notebooklm/

I am planning on automating upload of the final Markdown export to Google Drive.
#markdown #bookstack #wiki #llm #llms #notebooklm #export #webdev @bookstack

I know I’m supposed to hate #ai but is it possible to use AI to be an #anticapitalist ? #llm #llms #capitalism #anticapitalism #technofeudalism