The ‘information sources’ in the AI circle: just these 30 people. These 30 accounts power the global AI signal. Scrolling through 100 second-hand interpretations is worse than following 10 primary accounts. Skip the 100 hot takes. Follow the 10 people they’re quoting. This list covers 7 dimensions: Research · Engineering · Education · Open Source · Evals · AGI · Global Ecosystem Follow this list for a week, and your cognitive density will transform. Follow them for a week. Your AI literacy will compound.

━━━━━━━━━ Research · Top Minds ━━━━━━━━━

▍@karpathy – Often defines the LLM narrative months before everyone else. ▍@fchollet – Creator of Keras and ARC‑AGI. Deep thinking on what intelligence really is. ▍@ylecun – Deep learning pioneer, Meta’s Chief AI Scientist. Sharp critiques of mainstream paths. ▍@lilianweng – Former OpenAI, author of Lil’Log. The gold standard for technical deep dives. ▍@drfeifei – CV pioneer. Focused on human‑centered AI and spatial intelligence. ▍@demishassabis – CEO of Google DeepMind. The clearest window into the AGI roadmap. ▍@sama – OpenAI CEO. His posts often move the entire industry.

━━━━━━━━━ Engineering · Build & Ship ━━━━━━━━━

▍@simonw – Co‑creator of Django. The go‑to source for hands‑on LLM tooling and agents. ▍@rasbt – The benchmark for “build it from scratch” tutorials. ▍@swyx – Host of Latent Space. The one who put “AI Engineer” on the map. ▍@ggerganov – Creator of llama.cpp. You can’t talk local inference without him. ▍@hwchase17 – Founder of LangChain. First‑hand source on agent framework evolution. ▍@amasad – CEO of Replit. One of the sharpest voices on AI‑native developer tools.

━━━━━━━━━ Education · From Zero to Deep ━━━━━━━━━

▍@AndrewYNg – A legend in AI education. ML advice, courses, and real‑world deployment. ▍@jeremyphoward – Founder of fast.ai. Has spent years democratizing deep learning. ▍@maximelabonne – Fine‑tuning in practice. Books, code, all open‑sourced.

━━━━━━━━━ Open Source · Community Pulse ━━━━━━━━━

▍@_akhaliq – A live feed of arXiv papers, model drops, and open‑source projects. ▍@dair_ai – Frequent paper threads that make frontier research digestible. ▍@Teknium1 – Co‑founder of Nous Research. A leading voice in open‑source LLM training. ▍@ClementDelangue – CEO of Hugging Face. One of the highest‑quality signals in open‑source AI.

━━━━━━━━━ Evals · What Models Actually Do ━━━━━━━━━

▍@goodside – LLM evals, prompt research, real capability probes. Razor‑sharp observations. ▍@lmsysorg – The official Chatbot Arena. The most credible head‑to‑head model leaderboard. ▍@scaling01 – Scaling laws, model comparisons, independent benchmarks. A must for data nerds.

━━━━━━━━━ AGI · The Long Game ━━━━━━━━━

▍@ID_AA_Carmack – On AGI and low‑level optimization. Will rewire how you think about intelligence. ▍@gwern – Long‑form essays. AI research notes through a long‑term lens. ▍@jackclarkSF – Co‑founder of Anthropic, author of Import AI. Dense takes on policy and industry shifts. ▍@AnthropicAI – The official Anthropic account. Research cadence and safety work from the Claude team. ▍@ESYudkowsky – One of the most influential voices in alignment. Strong opinions, unavoidable reading.

━━━━━━━━━ China & Global Ecosystem ━━━━━━━━━

▍@teortaxesTex – The best non‑Chinese tracker of DeepSeek, Qwen, and the China AI scene. ▍@deepseek_ai – A flagship of Chinese open‑source LLMs. First‑hand model and research drops.

━━━━━━━━━ 15 → 30. This is a list of sources, not influencers. Not hype accounts, not KOLs, but the people actually producing first‑hand AI knowledge. Bookmark, repost, then follow them all. Who else should be added? Drop them in the comments.

fast.ai - Making neural nets uncool again

Source: https://www.fast.ai/

I Don’t Want a Learning Dashboard for My Child (https://www.fast.ai/posts/2026-02-17-education/index.html)education

technical

What analog and AI education both get wrong (https://www.fast.ai/posts/2026-02-17-education/index.html) Feb 17, 2026 Rachel Thomas (https://www.fast.ai/posts/2026-02-17-education/index.html)

Breaking the Spell of Vibe Coding (https://www.fast.ai/posts/2026-01-28-dark-flow/index.html)ai-in-society

technical

Sinister variations on the positive state of flow (https://www.fast.ai/posts/2026-01-28-dark-flow/index.html) Jan 28, 2026 Rachel Thomas (https://www.fast.ai/posts/2026-01-28-dark-flow/index.html)

How To Use AI for the Ancient Art of Close Reading (https://www.fast.ai/posts/2026-01-21-reading-LLMs/index.html)education

technical

Experiments in reading with LLMs (https://www.fast.ai/posts/2026-01-21-reading-LLMs/index.html) Jan 21, 2026 Rachel Thomas (https://www.fast.ai/posts/2026-01-21-reading-LLMs/index.html)

Stop Saying Boredom is Good for Kids (https://www.fast.ai/posts/2025-12-03-boredom/index.html)education

Chronic boredom causes stress, disengagement, and poor well-being in adults. So why do we glorify it for children? (https://www.fast.ai/posts/2025-12-03-boredom/index.html) Dec 3, 2025 Rachel Thomas (https://www.fast.ai/posts/2025-12-03-boredom/index.html)

A Guide to Solveit Features (https://www.fast.ai/posts/2025-11-07-solveit-features.html)technical

An overview of the features of the Solveit platform, which is designed to make exploration and iterative development easier and faster. (https://www.fast.ai/posts/2025-11-07-solveit-features.html) Nov 7, 2025 Solveit and Kerem Turgutlu (https://www.fast.ai/posts/2025-11-07-solveit-features.html)

Build to Last (https://www.fast.ai/posts/2025-10-30-build-to-last.html)advice

technical

Chris Lattner on software craftsmanship and AI (https://www.fast.ai/posts/2025-10-30-build-to-last.html) Oct 30, 2025 Jeremy Howard (https://www.fast.ai/posts/2025-10-30-build-to-last.html)

Let’s Build the GPT Tokenizer: A Complete Guide to Tokenization in LLMs (https://www.fast.ai/posts/2025-10-16-karpathy-tokenizers.html)technical

A text and code version of Karpathy’s famous tokenizer video. (https://www.fast.ai/posts/2025-10-16-karpathy-tokenizers.html) Oct 16, 2025 Andrej Karpathy, via Solveit and Kerem Turgutlu (https://www.fast.ai/posts/2025-10-16-karpathy-tokenizers.html)

How to Solve it With Code course now available (https://www.fast.ai/posts/2025-10-15-solveit2.html)courses

An email sent to all fast.ai forum users. (https://www.fast.ai/posts/2025-10-15-solveit2.html) Oct 15, 2025 Jeremy Howard (https://www.fast.ai/posts/2025-10-15-solveit2.html)

fasttransform: Reversible Pipelines Made Simple (https://www.fast.ai/posts/2025-02-20-fasttransform.html)technical

Introducing fasttransform, a Python library that makes data transformations reversible and extensible through the power of multiple dispatch. (https://www.fast.ai/posts/2025-02-20-fasttransform.html) Feb 20, 2025 Rens Dimmendaal, Hamel Husain, & Jeremy Howard (https://www.fast.ai/posts/2025-02-20-fasttransform.html)

What AI can tell us about microscope slides (https://www.fast.ai/posts/2025-02-03-cpath-foundation/index.html)health

technical

A friendly introduction to Foundation Models for Computational Pathology (https://www.fast.ai/posts/2025-02-03-cpath-foundation/index.html) Feb 3, 2025 Rachel Thomas (https://www.fast.ai/posts/2025-02-03-cpath-foundation/index.html)

A New Chapter for fast.ai: How To Solve It With Code (https://www.fast.ai/posts/2024-11-07-solveit.html)education

courses

fast.ai is joining Answer.AI, and we’re announcing a new kind of educational experience, ‘How To Solve It With Code’ (https://www.fast.ai/posts/2024-11-07-solveit.html) Nov 7, 2024 Jeremy Howard (https://www.fast.ai/posts/2024-11-07-solveit.html)

In defense of screen time (https://www.fast.ai/posts/2024-10-29-screen-time/index.html)education

Pundits say my husband and I are parenting wrong. (https://www.fast.ai/posts/2024-10-29-screen-time/index.html) Oct 29, 2024 Rachel Thomas (https://www.fast.ai/posts/2024-10-29-screen-time/index.html)

A new old kind of R&D lab (https://www.fast.ai/posts/2023-12-12-launch.html)ai-in-society

technical

Answer.AI is a new kind of AI R&D lab which creates practical end-user products based on foundational research breakthroughs. (https://www.fast.ai/posts/2023-12-12-launch.html) Dec 12, 2023 Jeremy Howard (https://www.fast.ai/posts/2023-12-12-launch.html)

Can LLMs learn from a single example? (https://www.fast.ai/posts/2023-09-04-learning-jumps/index.html)technical

We’ve noticed an unusual training pattern in fine-tuning LLMs. At first we thought it’s a bug, but now we think it shows LLMs can learn effectively from a single example. (https://www.fast.ai/posts/2023-09-04-learning-jumps/index.html) Sep 4, 2023 Jeremy Howard and Jonathan Whitaker (https://www.fast.ai/posts/2023-09-04-learning-jumps/index.html)

The real risk of AI is how it concentrates power (https://www.fast.ai/posts/2023-07-29-ai-centralizes-power/index.html)ai-in-society

Moving AI ethics beyond explainability and fairness to justice (https://www.fast.ai/posts/2023-07-29-ai-centralizes-power/index.html) Jul 29, 2023 Rachel Thomas (https://www.fast.ai/posts/2023-07-29-ai-centralizes-power/index.html)

AI Safety and the Age of Dislightenment (https://www.fast.ai/posts/2023-11-07-dislightenment.html)ai-in-society

Model licensing & surveillance will likely be counterproductive by concentrating power in unsustainable ways (https://www.fast.ai/posts/2023-11-07-dislightenment.html) Jul 10, 2023 Jeremy Howard (https://www.fast.ai/posts/2023-11-07-dislightenment.html)

Is Avoiding Extinction from AI Really an Urgent Priority? (https://www.fast.ai/posts/2023-05-31-extinction.html)ai-in-society

The history of technology suggests that the greatest risks come not from the tech, but from the people who control it (https://www.fast.ai/posts/2023-05-31-extinction.html) May 30, 2023 Seth Lazar, Jeremy Howard, & Arvind Narayanan (https://www.fast.ai/posts/2023-05-31-extinction.html)

Mojo may be the biggest programming language advance in decades (https://www.fast.ai/posts/2023-05-03-mojo-launch.html)technical

Mojo is a new programming language, based on Python, which fixes Python’s performance and deployment problems. (https://www.fast.ai/posts/2023-05-03-mojo-launch.html) May 4, 2023 Jeremy Howard (https://www.fast.ai/posts/2023-05-03-mojo-launch.html)

From Deep Learning Foundations to Stable Diffusion (https://www.fast.ai/posts/2023-04-04-part2-2023.html)courses

We’ve released our new course with over 30 hours of video content. (https://www.fast.ai/posts/2023-04-04-part2-2023.html) Apr 4, 2023 Jeremy Howard (https://www.fast.ai/posts/2023-04-04-part2-2023.html)

GPT 4 and the Uncharted Territories of Language (https://www.fast.ai/posts/2023-03-20-wittgenstein.html)ai-in-society

Language is a source of limitation and liberation. GPT 4 pushes this idea to the extreme by giving us access to unlimited language. (https://www.fast.ai/posts/2023-03-20-wittgenstein.html) Mar 20, 2023 Jeremy Howard (https://www.fast.ai/posts/2023-03-20-wittgenstein.html)

I was an AI researcher. Now, I am an immunology student. (https://www.fast.ai/posts/2023-02-07-school-immunology/index.html)education

Last year, I became captivated by a new topic in a way that I hadn’t felt since I first discovered machine learning (https://www.fast.ai/posts/2023-02-07-school-immunology/index.html) Feb 7, 2023 Rachel Thomas (https://www.fast.ai/posts/2023-02-07-school-immunology/index.html)

1st Two Lessons of From Deep Learning Foundations to Stable Diffusion (https://www.fast.ai/posts/2022-10-19-part2-2022-preview.html)courses

4 videos from Practical Deep Learning for Coders Part 2, 2022 have been released as a special early preview of the new course. (https://www.fast.ai/posts/2022-10-19-part2-2022-preview.html) Oct 19, 2022 Jeremy Howard (https://www.fast.ai/posts/2022-10-19-part2-2022-preview.html)

Deep Learning Foundations Signup, Open Source Scholarships, & More (https://www.fast.ai/posts/2022-09-23-part2-2022-signup.html)courses

Signups are now open for Practical Deep Learning for Coders Part 2, 2022. Scholarships are available for fast.ai community contributors, open source developers, and diversity scholars. (https://www.fast.ai/posts/2022-09-23-part2-2022-signup.html) Sep 23, 2022 Jeremy Howard (https://www.fast.ai/posts/2022-09-23-part2-