explore | Grilled Cheese

#LLMs are cool and good at generating lots of code. I'm sure there's more I need to learn about how I need to write #prompts or #requirements. As an example in the case of this financial news bot I'm writing: I should have specified the URL for the Tiingo documentation.

Neo4j Nodes 2026 is coming! November 12 2026. Virtual. 3 Timezones. 3 awesome agendas. Free, Virtual. Register today. Full agenda now available. #AI #Agents #Graph #LLMs #Memory #Reasoning neo4j.com/nodes/agenda/

Can you use the magic to defeat the magic?

Static testing breaks when #AIagents hit real users.

Zhou Yu shares how using AI agents to simulate complex user behaviors helps test your product automatically and at scale before deployment.

🎬 buff.ly/0E4c3qm

#AITesting #InfoQ #QConAI #LLMs

Common Crawl analyzed 584,107 llms.txt files and found most came from templates, many had no links, and some carried crawler rules the format can't enforce. via @mattgsouthern.bsky.social

#llms

Explore #AI, #LLMs, #coding, #robotics, #metaverse, & more at #EdTechKoreaFair 2026 in Seoul, Sept. 17–19!

Meet Kenji & Jolly at the #LPI booth to learn how #opensource skills prepare you for the future. Booth visitors get 20% off #LPI exams. 🐧💪 👇

lpi.org/jufh

#에듀테크코리아페어

1/🚨 New paper #emnlp2026

Do multilingual #LLMs work in English? It depends on how you ask. In "Lingua Franca or Probing Artifact?", we probe the same hidden states with three latent language identification methods and get different answers. 🧵👇

#interpretability #NLProc

OUT NOW:
Guying Zhou & Jiajin Xu present dassBNC2014: manually annotated corpus of directive speech acts in everyday British English conversations, thereby supporting investigations of the heterogeneity of directive practices in discourse.

#pragmatics #annotation #LLMs

doi.org/10.1075/ijcl...

"AI Safety" is mostly just corporate risk management dressed up as ethics. Most "refusals" aren't about preventing catastrophe; they're about avoiding the liability of a definitive answer. Are we building a god killer or just a very expensive autocomplete? #AISafety #LLMs

I'm going to speak at DevAI by Data Science Summit on the 19th of October!

My talk: "Harness Engineering - Principles and Practices"

Session is available as Online VoD - watch at your own pace.

#GenAI #LLMs #AI #AIAgents #AgenticAI #HarnessEngineering #GenAISDLC

#LLMs Are Generators – Do Not Forget That! - #ai

improveandrepeat.com/2026/09/llms...

Did you know that during the 2024 Red Teaming of Llama 3, researchers found you can force an LLM to break its system instructions by stacking conflicting adversarial prompts? It’s the digital equivalent of the "who's on first" routine. #LLMs #AI

#Bluesky really (really) hates #AI, #LLMs, #GPTs, #ChatGPT, #Claude, #Gemini, and more.

But when you have severe #ADHD, are #neurodivergent, or have executive function, cognitive, or mental deficits, that AI assistant can be a godsend.

Something to keep in mind as you lambaste data centers.

Should #AI still be benchmarked against humans?

Alex Gillespie’s #EHPS2026 keynote shared years of #research making sense of patient complaints with #LLMs.

One upshot: models often reveal what humans *miss* — not all human #alignment is good!

Follow on gScholar: scholar.google.com/citations?hl...

🔥 Can AI transform plastic waste into fuels & chemicals more intelligently?

This review explores how AI is reshaping plastic pyrolysis—from kinetic modeling and product prediction to interpretable AI, reactor optimization, and LLM-driven pyrolysis agents.

🔗 rdcu.be/QtEtz7nCPVqa

#AI #Energy #LLMs

#AnotherWordForCringe #AnotherNameFor

The C. D. #TheCD

1. Collective Distortion is another name for Large Language Model #LLM #LLMs

2. Collective Distortion = #LargeLanguageModels #LargeLanguageModel

3. #ChatGPT etc. Collective Distortion #MemeThinkingSystems

Year 1923 Finn's Hotel James Joyce.