All entries

the same model in a different harness can yield much different performance! we've seen this on a few different occasions now - we took gpt-5.2-codex from 52.8% to 66.5% on Terminal-Bench 2.0 (Top 30 t

The single most accurate science fiction author writing about AI turned out to be… Douglas Adams He wrote about AIs that work best when emotionally manipulated & that guilt you in turn. And he underst

The Top AI Papers of the Week (April 26 - May 3) - Latent Agents - RecursiveMAS - OneManCompany - AgenticQwen-30B-A3B - Agentic World Modeling - Agentic Harness Engineering - From Skill Text to Skill

This is a good explanation of why the gap between open and closed models is larger than it appears in benchmarks. I would add in that current open models are also more fragile than closed: they handle

To clarify, the current terminals can support peak download speeds of 400-500 Mbps. The 220 Mbps number is referring to actual in-flight performance given that the satellite to user beams are shared a

We have released a little skill to help you build LLM Wikis with your agents. Releasing a little skill to help build LLM Wikis. More people should be building LLM wikis. I hope the skill is useful for

This post from Harrison Chase (founder of LangChain) indicates that the LangChain team remains open to user feedback and continues iterating on their products and services based on community input. Th

When a software company actually uses its own software, all of their users benefit. @runwayml is exactly this type of company. They obviously take great care in the entire workflow that users desire.

The post humorously references Apple's iMessage feature that displays group chats in green when they include non-iPhone users, joking that activating 'goblin mode' (a playful term for behaving mischie

while exploring harnesses, i came across my first and favourite agent orchestration framework @LangChain has launched deep-agents last year. it’s a wonderful harness, that has a lot of tools: - write_

Why is the AI backlash growing? Outside of coding (where there is clear value), and a handful of other domains (e.g. brainstorming), Generative AI has been a net negative for society. GenAI has been u

This post shows a Wordle game result where the player solved puzzle #1,778 in 3 attempts, with the final answer being a five-letter word displayed in all green squares. The emoji grid represents the c

Yup The quiet shift no one is tracking: A year ago, defending free speech on X got you called a chaos agent. Now the same journalists demanding moderation are running their scoops on the platform firs

2022: “Stop overreacting, they won’t overturn Roe.” They did. 2023: “Stop overreacting, they won’t let women die rather than get an abortion.” They did. 2024: “Stop overreacting, they won’t arrest wom

3 things you can build for $0 on May 2nd: your website, your research, your internal systems. We’re celebrating 10 years of Replit by giving every user FREE access to Replit Agent for the day. Bring t

【革命】動画から3Dモーションを抽出！ComfyUIの新機能が凄すぎる 動画から人物の動きを抜き出し、高品質な3Dアニメーションに変換できる「Mocap Surgeon」が登場しました！✨ 注目の神機能はこちら： ・Jitter Filtering：動きのガタつきを数理的に除去し、滑らかさを実現 ・Slerp（球面線形補間）：関節のねじれを自然にブレンドして手動修正が可能 ・Time-Travel

5 months later, I would say that AI is definitely on track to produce 90% of all bullshit within a few years. Congratulations LLM Industry, you are changing the world! Claim AI will generate 90% of kn

Sam Altman expresses satisfaction with GPT-4.5's performance in high-speed mode, noting it delivers strong results. He reflects on being influenced by Twitter discourse on Medium regarding AI model ca

a bit more about why i think what Dawkins wrote is misguided: “Consciousness is not about what a creature says, but how it *feels*. And there is no reason to think that Claude feels anything at all. I

A lot of work around AI in 2023 was spent on building picks and shovels. I would know this because that was basically the core goal of the original @llama_index framework. Today, a lot of that is no l

This appears to be a humorous or personal post from Clem Delangue about merchandise or clothing items from Hugging Face, likely referencing company swag or branded apparel accumulated during his time

🚀 פרק חדש עם יובל פלג לוי מ @AI21Labs • למה חילוץ נתונים מ-PDF בעברית שובר גם מודלים מתקדמים? • הפתרון: החלפת מילים בעברית לאנגלית עם מידות פיזיות (Bounding Box) זהות, לשמירה על המבנה הוויזואלי מבלי ל

and see “Consciousness is not about what a creature says, but how it *feels*. And there is no reason to think that Claude feels anything at all. I am sure Claude can draw on its training data to wax p

and see also “Consciousness is not about what a creature says, but how it *feels*. And there is no reason to think that Claude feels anything at all. I am sure Claude can draw on its training data to

Are AIs about to subjugate humanity? In the debate about catastrophic AI risk, evolutionary scenarios receive far too little attention compared to largely speculative arguments about “instrumental con

The article discusses how the AI scaffolding layer—infrastructure tools and frameworks built around large language models—is experiencing consolidation and collapse, with LlamaIndex CEO explaining whi

Claude Opus 4.7 just implemented an AlphaZero-style self-play pipeline from scratch. It did this on consumer hardware in three hours, then beat the Pascal Pons solver 7 of 8 as first-mover on Connect

Composer 2 is 50% off in the SDK this weekend. Enjoy! We’re introducing the Cursor SDK so you can build agents with the same runtime, harness, and models that power Cursor. Run agents from CI/CD pipel

“Consciousness is not about what a creature says, but how it *feels*. And there is no reason to think that Claude feels anything at all. I am sure Claude can draw on its training data to wax poetic ab

// Contextual Agentic Memory is a Memo, not True Memory // Most agent memory today isn't memory. They are more like memos. A new paper argues that vector stores, RAG buffers, and scratchpads implement

This broadcast from Replit announced a countdown event leading to a 24-hour free agent period and the kickoff of a buildathon competition featuring over $100,000 in total prizes. The event likely focu

This post likely refers to an academic honor or recognition that Richard Dawkins received but Daniel Dennett did not, posted as a commentary by cognitive scientist Gary Marcus on X (formerly Twitter).

Gary Marcus comments on financial incentives in AI discourse, suggesting that posting sensationalized or false claims about artificial intelligence could generate significant revenue. This reflects Ma

For decades, prehistorical evidence was forced through a left-wing theory filter: primitive communism, the noble savage, and the myth that violence came from hierarchy, property, and “the system.” But

This Replit broadcast likely covered updates on Free Agent Day, a community event or initiative, and provided office hours for participants in a Buildathon (a hackathon-style building competition). Th

FSD Supervised can help you get over driving anxiety I had a patient today say to me that they wish they had a personal robot car to drive them around so they didn’t have to be stressed out in traffic

Generally, I would say X is not real life, but I am surprised about how often I get asked by executives about which AI lab is winning or what is up with a particular model in ways that indicate that t

great intro to agent harnesses and harness engineering! Is agent harness engineering the new prompt engineering? @LangChain co-founder and CEO, @hwchase17, joined us at #GoogleCloudNext to explain why

Grok 4.3 - excellent intelligence per unit cost xAI has launched Grok 4.3, achieving 53 on the Artificial Analysis Intelligence Index with improved agentic performance, ~40% lower input price, and ~60

Grok Voice is used by Starlink right now Grok Voice brutally dominates the top of the τ-voice Bench Grok scores 67.3%, while Gemini sits at 43.8% and GPT Realtime at 35.3% This is a massive lead over

This post showcases Golden Week (GW) travel in Toyama Prefecture, featuring fresh spring greenery in the foreground contrasted with snow-capped peaks of the Tateyama Mountain Range visible in the dist

here’s the essay: “Consciousness is not about what a creature says, but how it *feels*. And there is no reason to think that Claude feels anything at all. I am sure Claude can draw on its training dat

I added a new feature to my blog (built entirely on my phone with Claude code for web) that imports my iNaturalist photos and adds them to my site's overall timeline https://simonwillison.net/2026/May

Sam Altman reflects on the trade-off between model performance improvements and practical considerations like cost and speed, suggesting that increased intelligence remains a primary focus despite use

I was quoted a couple times in this Atlantic article, but that isn’t (the only) reason I think it is good. It lays out the reasons why we whipsawed from “AI is a bubble” to “there are not enough data

If only Richard Dawkins had seen this. 1/2 Why AI is unlikely to become conscious – my 2026 @TEDTalks is now online. What do you think about the prospects for 'conscious AI'? https://www.ted.com/talks