All entries

Openclaw token consumption dropped 50% within a month according to OpenRouter usage data, as reported by Gary Marcus. The post likely discusses potential causes for this significant decline, such as c

Our internal AI agent was a brute-force nightmare. 🛑 It burned 40,000 tokens, took 2 minutes, and only hit 68% accuracy just to answer a single question. Why? Traditional databases are built for human

Ethan Mollick discusses research findings showing that while improved prompting techniques provide significant benefits for AI model performance, the underlying model training remains the primary limi

Cursor's Bugbot tool detects code issues during development, with over 80% of default-effort issues being resolved by users before code merge. The high-effort detection mode identifies 35% additional

ComfyUI announced early access to a workflow publishing feature that allows users to share their AI image generation workflows on a hub or marketplace platform. This feature enables the community to d

Qwen 3.6 Plus by @Alibaba_Qwen is now FREE for a limited time on Nous Portal! Nous Portal is one easy subscription that gives you access to 300+ models, exclusive discounts, and bundles your tokens an

Clem Delangue announced the Reachy Mini robot is operational with audio capabilities enabled, with plans to integrate it with local AI services and a Hermes Agent in the near future. This indicates pr

Read the full blog: https://www.together.ai/blog/serving-deepseek-v4-why-million-token-context-is-an-inference-systems-problem# Watch the full webinar on DeepSeek v4: https://www.youtube.com/watch?v=D

real time model streaming audio/video/text in and out with tool use People talk, listen, watch, think, and collaborate at the same time, in real time. We've designed an AI that works with people the s

Really doubt what Hinton says here. Self-play for games like Go is not like the open-ended real world. Geoffrey Hinton says AI systems can keep improving when they are not limited by human data Like A

Gary Marcus analyzes and critiques Geoffrey Hinton's arguments regarding weaknesses in deep learning and artificial neural networks. The article likely examines specific technical or conceptual claims

reply to Hinton’s reply to me, for additional context: Dear @geoffreyhinton, I literally never said that AI systems “JUST regurgitate”; that’s plainly false. I don’t believe it, and I didn’t say it. (

Sarra builds apps with her two sons. Noni shipped Bamboo Brain into the App Store's top 12 in Education. Rebecca built the system she wished she'd had during her custody battle. To every mother buildi

Science publishing giant Elsevier has joined the dozens of firms and individuals suing artificial intelligence companies over their alleged use of copyrighted works in training AI models https://go.na

seems obvious now but 4 years ago i was getting blank stares when I talked about leveraging LLMs in VC Some ways to use GPT-3 as a VC: 📨Email/task triage - auto tag inbound, updates, etc. parse data a

Serious question: Should I write a short book called 7 lies about AI that never die? AI hype has become a giant game of bait and switch. the bait: we are going to make an AI that can solve any problem

Since Hinton has actually replied let me clarify some things - LLMS don’t *always* regurgitate - LLMs don’t literally store full texts - but given the mechanisms that they use they do sometimes regurg

🧵 Slime: The Most Elegant & Comfortable RL Training Framework Ever A deep dive into why Slime redefines LLM RL training with clean architecture & production-grade engineering ✨ Insights from Zhihu con

SpaceX has installed a cover on the Marmac 31 barge, a vessel designated to transport Super Heavy Boosters from Starbase to unspecified destinations. The barge, nicknamed 'You'll Thank Me Later,' repr

Sam Altman comments on OpenAI's new ChatGPT model, noting that the combination of improved capabilities, personality features, and personalization options has crossed an important threshold of functio

Starship’s cargo bay delivers 1,000+ cubic meters of usable volume The entire International Space Station has a pressurized volume of 1,005 m³ It took 42 assembly flights and over $150 billion across

Start 'claude agents' in a high level directory with all your repos in it (for me thats ~/Projects). It keeps track of which sessions need your input and makes it really easy to resume and pick up whe

Such hypocrisy Same move. Same hand on the heart, same wave to the crowd. Musk did it and gets called a Nazi to this day. Mamdani does it and those same people have not said a word. Well, not exactly.

Swap models & view their capabilities! Try out in Deep Agents CLI: https://docs.langchain.com/oss/python/deepagents/cli/ here's model profile details look like in practice, using @NVIDIAAIDev's Nemotr

That @NIST CAISI announcement from last week about Google DeepMind, xAI and Microsoft signing new deals for pre-deployment testing is gone from their website, link no longer found, and I can't get any

This post discusses strategies for scaling from managing a single AI agent to coordinating multiple agents efficiently, likely addressing workflow challenges and tooling improvements that eliminate th

arXiv:2605.06729v1 Announce Type: cross Abstract: We present the EDelta-MHC-Geo Transformer, a novel architecture that unifies Manifold-Constrained Hyper-Connections (mHC), Deep Delta Learning (DDL),

The inability of AI models to produce creative variation is a huge gap. The fact that they generate similar ideas limits their ability to do science & the same-y writing limits their usefulness in man

the inside story of the legendary Cog House. i believe there have not been any public photos of this place until now (bc we were explicitly not allowed to lol) as an advisor its been awe inspiring to

The internet gave language models their data for free. Robots don’t have that. Every trajectory has to be earned through hardware, time, teleoperators, and real consequences. Shrey’s piece on simulati

// The Memory Curse in LLM Agents // (bookmark it) Long histories apparently degrades agents as they become increasingly history-following and risk-minimizing. Across 7 LLMs and 4 social dilemma games

The next generation of models won't just generate images - they'll understand worlds, motion, interaction, and action. We've been building toward this for a while. Visual intelligence is becoming real

Thinky's secret plan: 1: Increase Human<->AI bandwidth 2: Raise ceiling of human+AI intelligence 3: Help humans continue as main-characters in the new world We are at Step 1. Interaction Models are gr

This post describes an AI system that monitors its own codebase to identify missing monitoring points and automatically generates pull requests to fix detected bugs, demonstrating autonomous self-impr

This is a tutorial on diffusion and flow matching, based on my previous postings here. I’ve made available the PDF, a python notebook for people to play with it, and the TEX source so hopefully one of

This is going to get even worse as people realize that careful tuning in their prompts can make AI writing seem not like AI writing to readers. We expect word counts to align, in some way, with thinki

This is harder to build than it looks. Preserving full conversational context while swapping underlying model providers mid-flight is a surprisingly deep systems problem. Most tools drop state or forc

Gary Marcus critiques the typical hype cycle pattern where emerging technologies experience inflated expectations followed by inevitable disappointment. The post likely illustrates this cycle using a

This seems like a critical reason to open up about AI use in academia. Scholars are using old AI models, badly, and not talking about it. New models hallucinate very few citations, and good agentic ha

This works really well btw, at the end of your query ask your LLM to 'structure your response as HTML', then view the generated file in your browser. I've also had some success asking the LLM to prese

To be clear, I meant 'idea' in the context of science and engineering, in particular AI, as in 'theory' or 'invention', like a new theory of cognition or an idea for a new ML approach. I thought that

To train better open models, we need predictable scaling. Delphi is Marin’s first step: we pretrained many small models with one recipe, then extrapolated 300× to predict a 25B-param / 600B-token run

Today, we're announcing $30M in new funding to build the AI OS for Research. 2.5M researchers start their work with Consensus every month. Their work is the foundation that all progress is built upon.

Today we’re launching the OpenAI Deployment Company to help businesses build and deploy AI. It's majority-owned and controlled by OpenAI. It brings together 19 leading investment firms, consultancies,

Today we're sharing our work on interaction models. A new class of model trained from scratch to handle real-time interaction natively, instead of gluing it onto a turn-based one. https://youtu.be/A12

This post from AI researcher Gary Marcus likely discusses the enormous economic value or potential return on investment related to artificial intelligence developments, suggesting AI's worth or impact