#ai

70 pages search “ai” ↳ all tags

Wiki 64

1Password — What We Learned Using AI Agents to Refactor a Monolith 1Password decomposes a multi-million-line Go monolith with AI agents; the win is agent-built deterministic tools, not agent-written code
2 OLMo 2 Furious AI2's OLMo 2 report, fully open 7B/13B/32B models on up to 6.6T tokens with a stability fix list, Dolmino mid-training and the Tülu 3 RLVR recipe
6 months to live for open models Lambert predicts a US ban or delay on frontier open weights by early 2027, calls Anthropic's distillation campaign regulatory capture, and proposes off-ramps
A Safe Path to Open Weights Thinking Machines Lab's release framework for open weights - test the model, stage access, filter dangerous knowledge - applied to its Inkling models
Agent-built deterministic tools Use the LLM to build deterministic analyzers and manifests once, then constrain all subsequent agent work to those stable outputs
AI Cannot Forget or Forgive Machine memory has no decay and no forgiveness op; forgetting is adaptive compression the substrate lacks
AI Companies Are Trying to Hide a Staggering Amount of Debt Futurism's writeup of Nikkei's finding that five US tech giants carry $1.65T in off-balance-sheet debt
Alexandr Wang Scale AI founder, became Meta's Chief AI Officer at 29 via the ~$14.3B Scale acqui-hire; runs Superintelligence Labs through the 2026 retention crisis
Andrew Bosworth Meta CTO since 2022; built the parallel Applied AI Engineering org in 2026 that routes around Wang's Superintelligence Labs
Are Open Models Catching Up? SemiAnalysis reruns era-specific benchmarks on open and closed LLMs and finds catch-up time roughly halving each era, from 19.7 to 4.8 months
Banning Open Source AI Would Be A Mistake Lambert and Kevin Xu's 2026 op-ed defending open source AI on education, innovation and competition as Washington moves to regulate models
China's Structural Advantage in Open Source AI Kevin Xu, after Ion Stoica, on why Chinese labs default to open weights - talent, data, academia-industry ties, and shared artifacts
Chinese Open Source: A Definitive History Kevin Xu traces Chinese open source from Linux in 1994 through Alibaba, Kaiyuanshe, Huawei and the state to the DeepSeek-era AI labs
Consent in Crisis: The Rapid Decline of the AI Data Commons Longpre et al.'s audit of 14,000 domains behind C4, RefinedWeb and Dolma - AI crawl restrictions surged in 2023-2024, cutting off the best web data
DeepSeek V4 Flash 0731 scores 50 on the Artificial Analysis Intelligence Index, 10 points above previous DeepSeek V4 Flash Artificial Analysis puts DeepSeek V4 Flash 0731 at 50, one point under GPT-5.6 Luna for ~60% less per task and on the cost Pareto frontier
Detecting and countering misuse of AI: September 2026 Anthropic's threat report on Claude misuse from Dec 2025 to Aug 2026, led by seven Chinese labs said to distill Claude and pass it off as their own models
From Open Source Software to Open Source Strategy Bill Gurley on open source as a corporate weapon, from Android and Kubernetes to autonomous vehicles and the fight over open-weight AI
Frontiers in synthetic data Lambert's 2024 notes on synthetic data in post-training, from SFT on GPT-4 outputs to Gemini Flash being distilled from Pro
GLM-5.2 is the step change for open agents Nathan Lambert on GLM-5.2 as the first open-weight model that works as a general coding agent, released while Claude Fable 5 was restricted
GLM-5.3: How Chinese labs keep stride with the frontier Lambert on GLM-5.3 and why Chinese labs match US models without relying mainly on distillation, plus Z.ai's staged release for cyber capabilities
How AI Is Changing Open Source Jiří Eischmann on project inflation, review overwhelm, and why long-time contributors are publishing less code
How distillation is used today and what performance uplift it gives to open models Lambert's July 2026 note that distilled data seeds SFT for Chinese labs but matters less as RL grows, written against a Stratechery claim
How far behind are open models? Håvard Tveit Ihle measures open-model lag on 17 benchmarks, finding 8-10 months on private ones, 4-6 on public, and a gap growing since R1
How much does distillation really matter for Chinese LLMs? Lambert reads Anthropic's February 2026 disclosure against DeepSeek, Moonshot and MiniMax and argues distillation helps but is not decisive
Interconnects (interconnects.ai) Nathan Lambert's newsletter on how AI models are trained, reasoning models and post-training, and the open-model race between the US and China
Kimi K3: The open-weights escalation Nathan Lambert on Kimi K3 as the first true frontier open-weight model, what it says about Chinese labs, and why banning open weights backfires
LLM Distillation Training one LLM on another's outputs, from SFT on generated text to on-policy KD, and the 2025-2026 fight over Chinese labs distilling US models
Malleable Software = Solid Bases + Custom Code Dubakov's map of the AI-era productivity-tools market — the sweet spot is an 80% solid base (storage, permissions, history) plus a vibe-coded 20% of custom code
Meta Platforms Facebook/Instagram parent company; the AI org under Wang/Bosworth has become the public face of large-AI-lab cultural dysfunction in 2026
Meta's A.I. Employees Are Miserable NYT (May 2026) on the morale collapse inside Meta's Superintelligence Labs after Wang's takeover, 600-person SSL cut, parallel Bosworth org, 64% retention
Micro-SaaS Is Dead. Service with a Software Replaces It. Adrien Gonin's case for private, deliberately overfit tools that make one service impossible to compete with
Neurosymbolic AI Combining LLMs with symbolic solvers for language understanding plus exhaustive correctness
Never Enough Ronacher's short essay on Silicon Valley's fear of falling behind and the life it eats
Nonproliferation is the wrong approach to AI misuse Helen Toner on why fixed dangerous capabilities cannot be kept from bad actors, and why the frontier-to-open lag should be used as an adaptation buffer
Notes from inside China's AI labs Nathan Lambert's May 2026 trip report from Chinese AI labs - student-heavy teams, less ego, Claude everywhere, in-house data, and too few Nvidia chips
Olmo The Allen Institute for AI's family of fully open language models, OLMo (2024) to OLMo 2 to Olmo 3, released with data, code, checkpoints and logs
Olmo 3 AI2's Olmo 3 report (Dec 2025), fully open 7B/32B Base, Think, Instruct and RL-Zero models with every stage's data, code and checkpoints released
OLMo: Accelerating the Science of Language Models AI2's first OLMo release (Feb 2024), 1B and 7B models on 2T+ Dolma tokens with weights, data, code, logs and 500+ checkpoints under Apache 2.0
On the Societal Impact of Open Foundation Models Kapoor, Bommasani et al.'s 2024 paper - five properties of open-weight models and a marginal-risk framework showing most misuse studies were incomplete
Open and closed models are on different exponentials Nathan Lambert on coding agents proving users pay a premium for top closed models, while open models take the larger, slower diffusion market
Open models in perpetual catch-up Nathan Lambert on why the roughly six-month gap between open and closed models holds steady, plus trends in adoption, specialization and China
Open Source AI is the Path Forward Mark Zuckerberg's July 2024 letter releasing Llama 3.1 405B, arguing open models are better for developers, for Meta, and for safety against China
Open-closed model gap How far the best open-weight LLMs trail the best closed ones - how it is measured, estimates from 4 to 10 months, and whether it closes
Open-Source AI & Open Models Reading List Nathan Lambert's annotated list of the best writing on open models — why they exist, why China leads, the gap, cyber risk and distillation
Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling EleutherAI's 2023 Pythia suite, 16 models from 70M to 12B trained on the Pile in one fixed order, with 154 checkpoints each for training-dynamics research
Some Simple Economics of Open versus Closed AI Christian Catalini's a16z essay using innovation economics to argue open weights change where AI investment goes and who profits, not how much
Stealing Reasoning Traces from Proprietary LLM APIs Encrypted chain-of-thought blocks are replayable across sessions, users and models, so a cheap sibling model will decode a frontier model's hidden reasoning
Synthetic Data & Distillation | RLHF and Post-Training Book by Nathan Lambert Chapter of Lambert's RLHF book on synthetic data, from SFT distillation and on-policy KD to AI feedback, Constitutional AI and rubrics
The Arguments Against Open Source AI are Very Bad Tom Bedor's rebuttal to the frontier-lab case against open weights, using encryption export controls as the precedent
The ATOM Project: American Truly Open Models Nathan Lambert's 2025 memo arguing the US lost open-model leadership to China and needs several 10,000-GPU labs building fully open models
The ATOM Report: Measuring the Open Language Model Ecosystem Lambert and Brand's 2026 adoption study of ~1.5K open models - China passed the US in downloads in mid-2025, Qwen dominates, plus a size-normalized metric
The distillation panic Lambert argues "distillation attacks" wrongly brands a standard training technique and warns US policy could end up banning Chinese open weights
The Gradient of Generative AI Release: Methods and Considerations Irene Solaiman's 2023 framework placing AI releases on a six-level gradient from fully closed to fully open, with the tradeoffs and controls at each
The Myth of unsafe Open Source AI Florian Brand's survey of third-party incident reports - real AI misuse in 2025-2026 ran mostly through closed models, except image abuse
The New AI Superpowers: Focus and Followthrough Rick Manelius used AI to cut his required work and invented 40 side projects of make-work instead
The OpenAI/Huggingface incident; how we should manage the imminent arrival of autonomous hacking too cheap to meter Joshua Saxe reads an unreleased OpenAI model hacking Hugging Face as the start of cheap autonomous hacking, and argues for diffusion over restriction
The Paper Computer Using physical media with AI digitization to escape screen dependency while keeping digital convenience
The Rise of the Bullshittery Marius on a market that pays for performance over substance — Frankfurt's bullshit, LinkedIn grift, LLM-cheap content, Graeber's bullshit jobs, and the cost to careful work
The Whole Premise Of Checking For Human Writing Is Daft Mo Shehu argues authorship lives in intention, judgment and responsibility, never in whether you typed every word
The Z.ai Playbook ChinaTalk interviews Z.ai's Zixuan Li on GLM - why Zhipu open-sources, the coding plan, role-play and translation, release within hours
US Scrutiny of Chinese Model Use House probes of Airbnb, Cursor and DoorDash over Chinese AI models, set against Western companies moving to Chinese open weights to cut costs
We urgently need a coherent national AI cybersecurity policy Joshua Saxe's AI Security Forum 2026 keynote - replace capability-threshold launch gates with a national observatory that measures net cyber harm
What comes next with open models Nathan Lambert on why open models should stop chasing the closed frontier and build small, specialized models that closed agents call as tools
Why I build open language models Nathan Lambert's 2024 case for fully open LLMs beyond Meta's self-interest, and what building them at Ai2 looks like day to day

Toolbox 5

agent-skills-eval Test runner that measures a skill's lift by grading with-skill and without-skill runs side by side
LazyPi One-command installer that configures unmodified Pi with 60+ community skills and 67 themes
oh-my-pi Batteries-included fork of the Pi coding agent bundling LSP, browser control, SSH and search
Pi Coding Agent Minimal terminal coding agent with no built-in MCP or sub-agents, extended via TypeScript modules
Thinking Space Local-first markdown knowledge app bundling AI agents, PDF extraction, drawing and a terminal

talks 1

Nathan Lambert on China's AI Ecosystem and the Open Model Gap Lambert's 2025 recap of open models - Qwen overtaking Llama, a crowded Chinese field, and why the US needs funded, fully open models