stacy @segvoid

On Friday someone asked me how they should think about what they should do with their career. I told her the most important thing to recognize is that progress proceeds as a series of S curves, and the most important thing is to recognize when an S curve is going to take off and get in at the bottom. The rest -- what exactly you do, for example -- is less important. This is true if you’re a scientist just as much as if you’re an entrepreneur. With AI today, we don't exactly know where we are on the S curve or when/if it will flatten, but we're certainly not exactly at the bottom anymore. That said, the economy is going to change dramatically in the next 10 years, and there are a bunch of other technologies that are clearly just over the horizon, so there are lots of other S curves that are taking off right now. There are major opportunities in every direction. There has probably never been a better time to be a scientist or entrepreneur.

Our paper on Robin is out at Nature! Robin was the first multiagent system for end-to-end biological research, which we preprinted last year, and was published back to back with Google's awesome Coscientist from @vivnat. Great validation, and major congratulations to the team.

Join us this Thursday to discuss how AI Scientists can empower scientific discovery with @scale_AI! Together with @GaoShanghua and @marinkazitnik, we will share our recent work on ClawInstitute, how AI Scientists can be built with ToolUniverse, and a sneaky preview of some new work AutoScientists. Much more to come! Link to register 👇

Biology needs AI people, just as much as AI needs biology people.

AI for biology is a lot harder than most pure tech people think. There are many bottlenecks, and high-quality data is the biggest one. If you actually wanna move the needle on AI/bio, then we need the following: - A major global project to expand high-quality biological data. - More foundation models for biology at the AlphaFold calibre. - A breakthrough in graph neural networks, as significant as transformers were for NLP. - New AI architectures that are first principles designed to natively work on bio problems. - A more intellectually honest and balanced research/business culture than the current hype game. - A major effort to identify, isolate, and exclude scientifically fraudulent papers and data points in biology. - Policy changes for more funding, cheaper electricity, targeted deregulation around the bioeconomy and use of anonymised data etc. - could be useful too.

This is the the quote I've been citing a lot recently.

kache @yacineMTB

5 months ago

you can outsource your thinking but you cannot outsource your understanding

287 4K 19K 2.9M 6K

Self-driving labs are transforming chemistry; but high cost & complexity limit access to a few well-funded labs. We wanted to change that. Our new paper in @NatureSynthesis introduces RoboChem-Flex 🧪🤖 🔗nature.com/articles/s4416… #selfdrivinglab #flowchemistry #optimization

I’ve wanted to do this for a decade. But I never did - I refuse to give any company my DNA. It is me. So this week I sequenced my genome entirely at home. Literally on my kitchen table. I never exposed my DNA sequence to the internet. Not at any point. I used a MinION to do the sequencing (it’s smaller + weighs less than an iPhone). I used open-source DNA models for the analysis (Evo2 and AlphaGenome) running locally on a DGX Spark and Mac Studio. I traced mechanisms behind my family’s multigenerational autoimmune conditions that no clinician has been able to understand. When I set out to do this I didn’t know if it would actually work. It does. Your genome is the most private data you will ever have. You probably shouldn’t let it leave your house.

keyboard_arrow_left Previous keyboard_arrow_right Next

Patrick Collison @patrickc

3 months ago

I'm lucky enough to have a great doctor and access to excellent Bay Area medical care. I've taken lots of standard screening tests over the years and have tried lots of "health tech" devices and tools. With all this said, by far the most useful preventative medical advice that

484 638 10K 4.1M 9K

Feel free to throw your hardest structural biology problems at Opus 4.7, and please share feedback so we can keep improving!

I’ve spent the past decade building bio AI models—until now, training them meant years, huge cost, and teams spanning AI, biology, and infra. Not anymore. Introducing a new capability at Biomni Lab: now any scientist can create, fine-tune, pre-train, and optimize bio foundation models on their own datasets, just by describing what they want. This is powered by a new feature called GPU-as-a-tool where we let AI agents launch and orchestrate GPU sandboxes. In the video, we show that you can use prompt to: - Fine-tune Borzoi for MPRA regulatory activity prediction - Fine-tune scGPT on a H1 hESC perturb-seq data - Fine-tune ESM2 for protein subcellular localization prediction - Pre-train a protein language model from scratch on UniRef - Build a novel multi-task ADMET model across 22 endpoints Another big challenge once you’ve trained or have access to a model is actually using it productively. As it is embedded within Biomni Lab, it bridges that gap, letting you “ask the model” to identify and prioritize relevant biological insights directly. Another exciting direction is lab-in-the-loop: scientists can design models, generate predictions, interpret results, and send them to the wet lab—all within one integrated biology environment. This is a preview capability and we’re looking for beta testers. Sign up here for early access: forms.gle/1yhCP6Vrc12DaS… Learn more about opportunities and limitations in our blog: phylo.bio/blog/ai-agents… @phylo_bio

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

A new paper published in Nature Astronomy says if LLM can easily replicate what counts as your scientific contribution, then the deeper problem is not the model, but the fact that the work was too routine, formulaic, or low-value to begin with. --- nature .com/articles/s41550-026-02837-2

I did this interview and kind of forgot. Just found it -they did a great job adding graphics and cutting it to make me sound smart. Amazing

TIME @TIME

4 months ago

In science, there's an "unlimited number of discoveries to be made." Edison's co-founder Andrew White says integrating AI could accelerate those possibilities.

123 262 2K 13.9M 523

🔥 AI just got its own infinite laboratory. Introducing LabWorld — the biggest leap yet for AI-powered science. LabOS just turned real biomedical protocols into fully executable, high-fidelity world model. This changes everything for AI scientists. 🧬⚡

Worth reading! technologyreview.com/2026/01/22/113…

I think this is one of the most important articles we've published at @AsimovPress. If you read carefully, there are at least 3-4 ideas in here that *should* be large, well-funded research programs. The article begins by arguing that existing AI models are good at predicting things *within* an existing framework, but are not good at building new frameworks (and, thus, cannot do paradigm-shifting science). As AI models become more widespread in science, they therefore risk "hypernormal science," meaning we will have less actual breakthroughs and more incremental discoveries. The author (Alvin Djajadikerta) supports this argument with several examples, one of which comes from germ theory: "In the mid-nineteenth century, doctors thought that illness was caused by noxious air, and kept meticulous records accordingly. The physician William Farr mapped cholera deaths across London and found they correlated strongly with low elevation, which he thought was because noxious vapors accumulated in low-lying areas. He was actually picking up a real signal: low-lying districts were closer to the contaminated Thames River. But because his data was organized around air quality, he could not find the true cause..." "An AI trained on Farr’s records could have found even subtler correlations, and would have been genuinely useful for predicting which neighborhoods would be hit hardest in the next outbreak. But it would not be able to derive the concept of a waterborne microorganism, as this was not a variable anyone had yet recorded." After giving other examples of this, Alvin begins mapping out ideas to solve this problem and create AIs that are "visionary" rather than "merely predictive." My favorite idea, of his, is to use AI agents as a model organism for metascience. The gist is that many paradigm shifts seem to happen under particular conditions. "Bell Labs, Xerox PARC, and the early Laboratory of Molecular Biology at Cambridge all produced extraordinary concentrations of paradigm-shifting work," Alvin writes, "mostly because they were small groups with enough institutional protection to pursue ideas that looked unproductive by conventional measures." Alvin continues: "We have never been able to run controlled experiments on scientific institutions; it is impossible to create labs that differ in only one respect and compare the results. But we could run AI agents in parallel populations under different research conditions, and analyze the results...In this sense, AI scientists may give metascience its first model organism." "For instance, one could test how group structure shapes discovery: do small, isolated teams produce more conceptual reorganization than large, well-connected ones? Do flat hierarchies outperform rigid ones? One could run AI agent populations that vary these factors independently and measure the results — something that is impractical to do with real institutions..." This essay is excellent throughout and I hope you'll read it.

keyboard_arrow_left Previous keyboard_arrow_right Next

We open sourced a version of the world's most capable AI co-scientist. Free. Easy to install. Has access to our Scientific Skills that are in use by 150k+ scientists worldwide. Please star the GitHub repo and repost/retweet.

K-Dense @k_dense_ai

4 months ago

We just open-sourced K-Dense BYOK, your own AI research assistant, running locally with your API keys. 170+ scientific skills. 250+ databases. 40+ models. Scalable compute via @modal when you need it. No subscriptions. No lock-in. Data stays on your computer. Repost, star and try

3 101 657 146K 1K

Holy shit... AI just learned scientific taste 🤯 For 400 years, one thing separated great scientists from average ones. Not intelligence. Not work ethic. Taste. The ability to look at 1,000 research directions and know which one is worth your life's work. Fudan University just bottled that into a model. They scraped 2.1 million arXiv papers and did something nobody had tried before. Instead of training AI to run experiments or search literature, they trained it to judge ideas. 700,000 matched paper pairs. Same field. Same year. Different citation counts. One job: figure out which research the scientific community actually cared about. They called it Scientific Judge. And the numbers destroyed every benchmark. It beats GPT-5.2, Gemini 3 Pro, and GLM-5 at predicting scientific impact. It generalizes to papers published after its training data ends. It works on fields it was never trained on. It even transfers to ICLR peer review scores without ever seeing them. But here's the part that broke my brain. They used Scientific Judge as a reward model to train a second AI called Scientific Thinker. You give it a paper, it reads it, then proposes the next high-impact research direction. Not a summary. Not a literature review. An original idea. Win rate against baseline? 81.5%. Win rate against GPT-5.2 itself? 54.2%. The AI is now proposing research ideas that frontier models judge as better than their own. Scientific taste was always the last human advantage in research. The PhD took 5-7 years because that's how long it took to develop judgment — to know what matters before anyone else does. That just became a training objective. Not from human feedback. Not from expensive annotations. From citations. The raw signal of what the scientific community collectively decided was worth building on. We're not talking about AI that executes science. We're talking about AI that decides what science is worth doing. That's a completely different thing.

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.

my hopes and dreams of working at Edison have been shattered (competition is now higher than ever)

Sam Rodriques @SGRodriques

4 months ago

Earlier this week at GTC, we announced our partnership with Nvidia. We will work with Nvidia to build strong, American open-source models that are at the frontier of scientific reasoning. These models will be essential for the US to compete with China on science in the coming

8 22 131 23K 43