- wöchentlichErscheinungsrhythmus
- 12 mintypische Länge
- vor 5 Tagenletzte Folge
Folgenzu den neuesten
The White House's plan for open models & AI research in the U.S.
Today, the White House released its AI Action Plan (https://www.ai.gov/action-plan), the document we’ve been waiting for to understand how the new administration plans to achieve “global dominance in artificial intelligence (AI).” There’s a
Kimi K2 and when "DeepSeek Moments" become normal
https://www.interconnects.ai/p/kimi-k2-and-when-deepseek-moments The DeepSeek R1 release earlier this year was more of a prequel than a one-off fluke in the trajectory of AI. Last week, a Chinese startup named Moonshot AI dropped Kimi K2 (h
The American DeepSeek Project
https://www.interconnects.ai/p/the-american-deepseek-project While America has the best AI models in Gemini, Claude, o3, etc. and the best infrastructure with Nvidia it’s rapidly losing its influence over the future directions of AI that un
Some ideas for what comes next (Jun. 2025)
https://www.interconnects.ai/p/summertime-outlook-o3s-novelty-coming Summer is always a slow time for the tech industry. OpenAI seems fully in line with this, with their open model “[taking] a little more time (https://x.com/sama/status/193
Crafting a good (reasoning) model
Why are some models that are totally exceptional on every benchmark a total flop in normal use? This is a question I was hinting at in my post (https://www.interconnects.ai/p/sycophancy-and-the-art-of-the-model) on GPT-4o’s sycophancy, wher
The rise of reasoning machines
https://www.interconnects.ai/p/the-rise-of-reasoning-machines Note: voiceover coming later in the day. I may fix a couple typos then too. A sufficiently general definition of reasoning I’ve been using is: Reasoning is the process of drawing
What comes next with reinforcement learning
https://www.interconnects.ai/p/what-comes-next-with-reinforcement First, some housekeeping. The blog’s paid discord (access or upgrade here (https://www.interconnects.ai/p/discord)) has been very active and high-quality recently, especially
How I Write
https://www.interconnects.ai/p/how-i-write My experience with my recent years of writing is quite confusing — almost even dissociative. I've never felt like I was a good writer and no one really told me I was until some random point in time
A taxonomy for next-generation reasoning models
https://www.interconnects.ai/p/next-gen-reasoners On Monday of this week we released RewardBench 2, Ai2’s next reward model evaluation and a project I’ve been personally invested in through its whole arc. Read more of my thoughts here (http
Claude 4 and Anthropic's bet on code
https://www.interconnects.ai/p/claude-4-and-anthropics-bet-on-code Claude’s distinctive characteristics are having a best-in-class personality and the ability to effectively perform software engineering tasks. These characteristics both app
People use AI more than you think
https://www.interconnects.ai/p/people-use-ai-more-than-you-think I was on ChinaTalk again (https://open.spotify.com/episode/5yPJwQkzhRPuEsLHuZTq1o) recently to talk through some of my recent pieces and their corresponding happenings in AI.
My path into AI
https://www.interconnects.ai/p/how-i-got-here Some longer housekeeping notes this week: * I wrote briefly (https://natolambert.substack.com/p/on-the-new-openmdw-open-source-ai) about a new open-source license, OpenMDW from the Linux Foundat
What people get wrong about the leading Chinese open models: Adoption and censorship
https://www.interconnects.ai/p/what-people-get-wrong-about-the-leading Two editor’s notes to start. * First, we released our OLMo 2 1B model (https://huggingface.co/allenai/OLMo-2-0425-1B-Instruct) last week and it’s competitive with Gemmas
State of play of AI progress (and related brakes on an intelligence explosion)
https://www.interconnects.ai/p/brakes-on-an-intelligence-explosion Intelligence explosions are far from a new idea in the technological discourse. They’re a natural thought experiment that follows from the question: What if progress keeps g
Transparency and (shifting) priority stacks
https://www.interconnects.ai/p/transparency-and-shifting-priority The fact that we get new AI model launches from multiple labs detailing their performance on complex and shared benchmarks is an anomaly in the history of technology products
OpenAI's o3: Over-optimization is back and weirder than ever
https://www.interconnects.ai/p/openais-o3-over-optimization-is-back Over-optimization (https://rlhfbook.com/c/17-over-optimization.html) is a classic problem to reinforcement learning (RL) proper, the RL from human feedback (RLHF) that gave
OpenAI's GPT-4.1 and separating the API from ChatGPT
https://www.interconnects.ai/p/openais-gpt-41-and-separating-the Recently I gave another talk on RLVR (https://www.youtube.com/watch?v=zYeIqzULzr0) experiments and I posted some thoughtson OLMoTrace (https://natolambert.substack.com/p/looki
Llama 4: Did Meta just push the panic button?
https://www.interconnects.ai/p/llama-4 Where Llama 2’s and Llama 3’s releases were arguably some of the top few events in AI for their respective release years, Llama 4 feels entirely lost. Meta has attempted to reinvent their formula of mo
RL backlog: OpenAI's many RLs, clarifying distillation, and latent reasoning
https://www.interconnects.ai/p/rl-backlog-openais-many-rls-clarifying I have a second blog where I post half-baked thoughts, sometimes previews of what comes here. If you’re interested, I posted some musings on OpenAI’s coming open model re
Gemini 2.5 Pro and Google's second chance with AI
https://www.interconnects.ai/p/gemini-25-pro-googles-second-ai-chance Google, with its immense infrastructure and talent, has been the safe bet for the question of “Who will have the best models in a few years?” Google took a long time to g
Managing frontier model training organizations (or teams)
https://www.interconnects.ai/p/how-to-manage-ai-training-organizations It is a closely guarded secret how the leading AI laboratories structure their training teams. As with other technology companies, the saying “you ship your org chart” s
Gemma 3, OLMo 2 32B, and the growing potential of open-source AI
Post: https://www.interconnects.ai/p/gemma-3-olmo-2-32b-and-the-growing Ever since the release of the original ChatGPT, much has been said about making a truly open-source version of it — with data, code, weights, etc., all available. Open-
Interviewing Eugene Vinitsky on self-play for self-driving and what else people do with RL
Eugene Vinitsky (https://www.eugenevinitsky.com/) is a professor a New York University department of Civil and Urban Engineering. He’s one of my original reinforcement learning friends from when we were both doing our Ph.D.’s in RL at UC Be
Elicitation, the simplest way to understand post-training
Full post: https://www.interconnects.ai/p/elicitation-theory-of-post-training If you look at most of the models we've received from OpenAI, Anthropic, and Google in the last 18 months, you'll hear a lot of "Most of the improvements were in
Where inference-time scaling pushes the market for AI companies
Link: https://www.interconnects.ai/p/where-inference-time-scaling-pushes There’s a lot of noise about the current costs of AI models served for free users, mostly saying it’s unsustainable and making the space narrow for those with the hist
GPT-4.5: "Not a frontier model"?
More: https://www.interconnects.ai/p/gpt-45-not-a-frontier-model As GPT-4.5 was being released, the first material the public got access to was OpenAI’s system card for the model that details some capability evaluations and mostly safety es
Character training: Understanding and crafting a language model's personality
https://www.interconnects.ai/p/character-training The vast majority of evaluations used to measure progress on post-training at frontier laboratories are internal evaluations rather than the evaluations you hear about all the time like MATH
Claude 3.7 thonks and what's next for inference-time scaling
On Monday, February 24th, 2025, Anthropic announced their latest model (https://www.anthropic.com/news/claude-3-7-sonnet), Claude 3.7 Sonnet, which is their first model explicitly trained to use more inference time tokens to improve perform
Grok 3 and an accelerating AI roadmap
Full post: https://www.interconnects.ai/p/grok-3-and-an-accelerating-ai-roadmap xAI launched their latest flagship model, Grok 3, last night via a live stream on X (https://x.com/xai/status/1891699715298730482), which is a new take on the l
An unexpected RL Renaissance
The era we are living through in language modeling research is one characterized by complete faith that reasoning and new reinforcement learning (RL) training methods will work. This is well-founded. A day (https://x.com/jonasgeiping/status
