My 6200 Likes & Retweets
A curated list of liked, bookmarked, and retweeted posts from Twitter.
🧵 Today we’re publishing our first Research Agenda – a detailed outline of the most urgent questions we’re working to answer as AI capabilities grow. It’s our roadmap for tackling the hardest technical challenges in AI security.
I’ve left OpenAI! Already miss everyone on the Training team & my friends ❤️ but very excited to soon announce what’s next Until then, I’ll be taking a break to solve OCR for Sanskrit so we can immortalize the classical Indian literary canon in the weights of superintelligence https://t.co/A3iGgMl6WT
“I’m not religious, I believe in science” Okay… but… “The more I study science, the more I believe in God.” - Albert Einstein “In the absence of any other proof, the thumb alone would convince me of God’s existence.” - Isaac Newton “Both religion and science require a…
This is why it's important to surround yourself with people who can see the real you. If not, you might find yourself morphing your personality to match people's (incorrect) model of you, subconsciously hiding your real self away. https://t.co/ydGII2LAjI
Introducing Qwen3! We release and open-weight Qwen3, our latest large language models, including 2 MoE models and 6 dense models, ranging from 0.6B to 235B. Our flagship model, Qwen3-235B-A22B, achieves competitive results in benchmark evaluations of coding, math, general… https://t.co/JWZkJeHWhC
Someone made a helpful guide to accessing scientific papers: https://t.co/FJ4GXlgLPM https://t.co/CN5YcxMMTR
How can humans realistically travel to another star, and why will it be an all-female crew that does it? In this thread: Sailing on light, nuclear pulses, using the sun as a telescope and how to travel to another solar system. The interstellar thread! https://t.co/fNr1aku1bF
Project DeepWiki Up-to-date documentation you can talk to, for every repo in the world. Think Deep Research for GitHub – powered by Devin. It’s free for open-source, no sign-up! Visit deepwiki com or just swap github → deepwiki on any repo URL: https://t.co/5bHbvq98Ud
When we began fueling old language models with massive amounts of modern GPUs, some people briefly thought we had discovered true Artificial Intelligence. What we actually discovered was Natural Stupidity. Up until this point in history, we assumed that almost all humans,… https://t.co/1YOH5EyibL
the 3 AI paths you can go down right now... - creative (text, image, video, sound) - agents/workflow-automation - coding/development i recommend acquiring a baseline understanding of each path before diving deep into one
🚨 New AISI research 🚨 RepliBench is a novel benchmark that measures the ability of frontier AI systems to autonomously replicate. Read the full blog here: https://t.co/ewFJTytscV https://t.co/TCAb4HBCNe
Two undergrads. One still in the military. Zero funding. One ridiculous goal: build a TTS model that rivals NotebookLM Podcast, ElevenLabs Studio, and Sesame CSM. Somehow… we pulled it off. Here’s how 👇 https://t.co/8cfJSegciX
my favorite Jorge Luis Borges anecdote is that he hated thick books and thought you should just pretend the book you’re writing has already been written and instead write a summary of it https://t.co/d7g23eW8WS
Hard times creates manufacturing men Manufacturing men create higher GDP Higher GDP creates Saas men Saas men create hard times
"Increasingly, people don't become academics because of intelligence, rather because of lower grasp of disorder." - Nassim Nicholas Taleb
https://t.co/JcbJPTuZvj https://t.co/ZY8LboGD59
I need to combine 5am flow state with 11pm self-esteem with 2am creativity with 8am clarity with 11am optimism with 4:55pm vibes
Hi @Shopify, room for one more? https://t.co/gUVID18s6G
Vercel's State of AI Developer Survey results are in. Our report covers: • What teams are building with AI • LLM provider adoption • Top technical challenges in building AI apps • How companies are budgeting and staffing AI teams • AI trend changes from Q4 2024 to Q1 2025 https://t.co/tsmhVtRl1p
Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation. https://t.co/rDaqV0x0wE
It's time to set a Guinness World Record. Registration is now LIVE at https://t.co/y3OwMLa1N1: → $1M+ in prizes → Global IRL events → Legendary founders & investors judging If you've got ideas for apps or startups, this is your moment— *especially* if you don't code👇 https://t.co/BnPdIiYkF4
I have zero respect for a researcher, perched in an ivory tower publishing thousands of papers, who publicly tells a practicing clinician, who spends their day caring for patients, that they are a "nobody" because they have not published a paper.
Today, we are releasing the first version of Grok studio, adding code execution and google drive support. Grok Studio Grok can now generate documents, code, reports, and browser games. Grok Studio will open your content in a separate window, allowing both you and Grok to… https://t.co/lyQh06F8eP
our logo is a prominent aspect of our brand—one that appears in many of our designs. it's a symbol with layers of meaning, each one representing our core mission. https://t.co/VgA7M6EOue
There is a new platform, Sci-Net (https://t.co/SPy41cSJP5) where researchers can help each other by providing paywalled papers. The platform builds upon Sci-Hub meme token $scihub to reward active members. https://t.co/6lEFFyM0Yf
For decades, the startup world has pushed one narrative: you need co-founders to succeed. What if they’re wrong? The case for @solofounding: https://t.co/uDJB6rwwIs
No. Physics is about finding the shortest *encoding function* that maps from concrete physical states of the universe onto abstract representational states of our internal models. To see trivially why @BasedBeffJezos's definition doesn't work... (1/2) https://t.co/bIUQGZ9lGr
Season 1 begins today. 25 builders. 30 days. One shot to build and launch. https://t.co/wYgAdgzU8y
The best feature of LLMs is the shamelessness with which we can ask the dumbest questions. Which is the prerequisite to intellectual and personal growth.
Now everybody thinks browsers in the cloud is cool. Cool.
This list is automatically updated weekly (latest: 14 May 2026).