interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/

OPD forgets less than SFT, yet breaks under teacher mismatch and on long rollouts. Why? Our answer: they optimize different targets. With multiple teachers, SFT (forward KL) learns their arithmetic mixture; OPD (reverse KL) learns their normalized geometric mean. 🧵

AI Drug Discovery Is Becoming a Bottleneck Trade I get excited when an industry starts going through a real regime change. AI drug discovery (“AIDD”) increasingly looks like one of those moments. Two things are happening: 1/ Upstream: the frontier AI labs are piling in. 2/ Downstream: The supply chain is clearly moving. Supply-chain checks suggest the upstream picks-and-shovels of discovery and early preclinical R&D are starting to feel the increase in experimental volume. DNA → protein → assays → sequencing → automation → preclinical testing Names across that stack include $TWST , @GenScript , $ILMN , $TXG , lab-automation vendors and CROs. $TWST expects triple-digit percentage growth in AI-enabled drug-discovery orders in FY26, and another year of triple-digit order growth in FY27. @GenScript's AIDD business doubled YoY in 1H26. Its current platform advertises industrial-scale validation of 4,000+ designs/day, with integrated sequence-to-data workflows. Our channel checks suggest the ramp is moving even faster: roughly 8,000 designs/day currently, with a path toward ~16,000/day by YE26. --- Why does AI drive more wet-lab demand? 1/ AI makes hypothesis generation almost free → way more shots on goal. The bottleneck is moving from expert-driven design to biological validation. 2/ AI models need continuous experimental feedback — and both good and bad data are useful. Traditionally, only the highest-conviction A+ candidates might get pushed into expensive validation. With AI, even the B/C candidates can be valuable because failed experiments generate training data. @GenScript has said its sequence-to-binding workflow can return data in 4–7 days, and that faster cycle times matter because AI models depend on continuous experimental feedback. @Anthropic is a clean example. @claudeai designed 1,320 protein binders. @adaptyvbio converted those digital sequences into DNA, expressed the proteins and tested binding. Only 354 actually bound. And the 966 failures are not wasted. They are useful negative labels: what does not express, what does not bind, what has poor affinity. Those results help train the next model iteration. 3/ Wet labs are no longer just making drugs. They are making training data. $TWST / @GenScript increasingly look like biological data foundries. $TWST explicitly talks about generating model-ready data from AI-designed sequences. In some workflows, the customer may care less about receiving the physical protein than about getting structured experimental results back into the model. Traditional drug discovery asks: “Does candidate X work?” AI drug discovery also asks: “What can this experiment teach the model?” --- TAM of AIDD If AI is simply a better R&D tool, the relevant spending pool is the $300–400B of annual global pharma R&D. If AI meaningfully increases the number of viable drug programs, the opportunity is larger because it expands downstream demand for DNA synthesis, protein production, assays, and preclinical work. Near term, we can also size demand from AI-company spending. If Anthropic reaches $80B of ARR in 2026 and spends just 1% on AIDD, that alone would imply ~$800M of annual investment. --- Trade setup This is a trade that could have long legs. It’s hard to really stop working until PhaseI/II results (2028+) It smells a lot like the bottleneck trade we just saw in semis: GPUs → HBM → networking → power/cooling. In biology, It basically follows the drug discovery process downstream: AI models → designs → DNA/protein → assays → preclinical capacity. After the upstream picks-and-shovels, animal testing could become the next bottleneck. Monkey prices are already near prior highs and CRO capacity is tight. AIDD pushing more candidates into preclinical development would only add demand. ?? But clinical trials are still the bottleneck? This is the biggest pushback I keep coming back to. No matter how fast discovery becomes, drugs still need to go through preclinical → Phase I → Phase II → Phase III → approval. You still need patients, time and capital. But that doesn’t mean the bottleneck trade won’t work. More viable candidates — especially with higher success rates — still means more demand throughout the development process. And who knows: clinical trials themselves may eventually be optimized by AI. ?? What breaks the trade? Near term, the picks-and-shovels trade breaks if experimental budgets stop growing, AI-generated designs don’t translate into useful wet-lab hits, or capacity catches up too quickly. Longer term, the thesis breaks if AI drugs look great in discovery / Phase I but fail at normal rates in Phase II/III. That is why Phase II matters so much. ?? Milestones Late 2026–2027: first Isomorphic-designed drugs enter human trials; more AI-native programs move into IND-enabling work / tox. 2028–2030: clinical trial results start telling us whether AI-designed drugs actually perform better than conventional drugs. Calling all the "bottleneck bros". :) @jukan05 @zephyr_z9 @aleabitoreddit @ParadisLabs +++ More comprehensive analysis: robonomics.substack.com/p/ai…

Today, we’re launching the Geodesic API Platform. geodesiclab.com/api For the first time, Geodesic is making our AI models and drug discovery capabilities available through APIs. At its core is NovaAtom, our in-house model for predicting the structures of proteins, small molecules, nucleic acids, and their interactions. Now available in Lite Preview, this is the first time NovaAtom has been made available beyond Geodesic’s own discovery platform. Alongside NovaAtom, we’ve integrated the field’s core models — carefully curated and continuously updated. To use the API: curl -X POST platform.geodesiclab.com/api… \ -H "Authorization: Bearer $GEODESIC_API_KEY" \ -F "model=novaatom" \ -F 'params={"jobName":"rbd-nanobody","sequences":">rbd\nTNLCPFGEVFNATRFASV...\n>nb\nQVQLVESGGGLVQAGGSLR..."}' Poll the job, pull the results, and feed them directly into the next model. Your AI assistant can use the platform directly, too. Connect Claude Code, Codex, or your own agent with one URL and an API key. We manage the integrations and handoffs. You focus on the science. Understand biology. Predict structures. Design proteins. Discover molecules. One key. One interface. Start building today!

NovaDDE Plus is now live! Built for scientists who use AI for drug discovery every day — more compute, priority access, concurrent runs, API access, and end-to-end campaign execution. $30/month and sign up by September 25 to get your first month free. Our goal is simple: put the most powerful AI for drug discovery into the hands of every scientist. geodesiclab.com/pricing

We’re launching the Geodesic Grand Challenges. What are the hardest questions in drug discovery that, if solved, could change what’s possible over the next 5–10 years? We’re asking you to help define them. Selected questions will become official Geodesic Grand Challenges. Each selected contributor receives 10,000 Geodesic credits. Once the challenges are selected, we’ll open a call for solutions, with a $1 million prize for a verified solution to each challenge. The next breakthrough starts with the right question. Submit yours！ geodesiclab.com/grand-challe…

We built high-throughput materials labs in Menlo Park to create a loop between experiments and models. The labs generate fresh data, the models learn from it, and then help us decide what to try next. Using only 1,300 H200s, plus months of our experimental data, we mid-trained and RL’d an open-source model to surpass GPT-6 Astra on our analysis benchmark. We call it Neon. This is real footage from our lab. We’re focusing first on hard problems in materials science, including superconductors, magnets, and semiconductor materials. Read our blog posts below.

25 Fields Medalists are worried AI solving math too fast could damage mathematics. I see the opposite happening in coding. Imagine scientists delaying a cancer cure by 100 years just so they can experience discovering it themselves. It would be a disaster for humanity.

Here is a potential roadmap to reversing aging by 2040, created by GPT-6 Pro with Image 2.5. It is important to point out that we will already reach longevity escape velocity by around 2035, so if you can make it to then, you are very likely to live for hundreds of more years!