research lead @ openai
sf
Joined March 2009
- Tweets2.4K
- Following2.4K
- Followers10.8K
- Likes23.4K
rapha retweeted
We started Arda in 2025 to speed up American manufacturing. We have spent the last year building with Anduril as our design partner. Today we are launching the company publicly on CNBC and announcing our Series A: $160M at a $1.5B valuation.
Our platform helps manufacturers design and operate manual assembly lines, putting designs into production faster now. Next we'll build automated lines that can be reconfigured as quickly as human workers, letting America make products that now require armies of workers in Asia.
rapha retweeted
the Lord came down to see Adam's neural networks. in their greed, they had embedded more features than residual dims.
and the Lord spake, "let us confuse their activations, lest nothing they plan be impossible." thus the Lord scattered the features over the residual stream
rapha retweeted
The mood has been a bit dour lately, so here are some less-obvious things that I expect to take shape over the next decade that I am excited about:
1. Space! Lots and lots of stuff is going to be happening in space by the mid-2030s, and right now the U.S. looks like it will be the leader by default, modulo the overhyped “race” to land a human on the moon again. Robots on the moon building and digging and cutting and adhering are more exciting than a person standing there taking pictures. The humans can come once the machines have built us habitats.
2. Increasing substitutability of hydrocarbons. This will lead to fewer supply chain disruptions and more stable, lower prices. This is a story about both supply elasticity and demand elasticity, and the geopolitical implications are underrated. Amazing for North America.
3. Private market exposure increasingly being an option for public investors, combined with Silicon Valley figuring how to use debt and other financial instruments more creatively than in the past. This seems like something America will be uniquely good at and I expect it to boost our already enviable lead in emerging technology capital allocation.
4. State and local government adoption of AI. This will vary widely by jurisdiction, but in at least some places you will see government administrative processes and public service provision have huge increases in efficiency. State and local government is where most of the American public services that people rely on are actually administered, so this is high-impact for people’s lives.
5. Advances in mathematics rapidly diffused to other fields. Historically it takes on the order of decades or centuries for discoveries from the bleeding edge of mathematics to be translated into practical applications, but I think AI will accelerate this dramatically. Lots of this will have the flavor of “small steps toward a much better world,” but it will accumulate.
6. Speaking of mathematics, lean! Lean is a programming language used by mathematicians to formally guarantee the truth of a proof. AIs can write lean and eventually they will be superhuman at it. This will have applications far beyond math, including cybersecurity and contract writing/negotiation/execution.
7. Fixing American’s shipbuilding problem by building boats instead. America has great boat-building capacity, but is awful at building ships. I want us to get better at building ships, but I am confident we can scale boat production. The Navy will soon have massive fleets of autonomous boats everywhere in the world.
8. New kinds of agriculture will become possible because of autonomy. Row-crop agriculture is heavily mechanized already, and automating the tractor, by itself, does less than you think for the economics of a farm. But with time, more labor-intensive crops will become economical. And vastly more diverse and complex crop arrangements will become possible too, which will be both ecologically, economically, and culinarily beneficial. The economics of agriculture could start to be kinder to small farms again.
9. Because of the previous item and advancements in biotechnology, logistics, and robotics, I expect that people are underrating how much healthier, more diverse, and flat-out better the American diet will be in 2035.
10. The subterranean will become an increasingly important field of human activity. New imaging sensors, including ones enhanced by an abundance of space-based imaging platforms, will make the subterranean more immediately legible. Better drilling and digging will become cheap and widely diffused. Among many other things, the opening up of the subterranean as an industrial frontier will change what we think of as a usable natural resource.
11. Greater variety and ornamentation in the physical world. Standardization is partially a function of scarce labor. Robotics can make beauty accessible again, if we choose to let it.
There are many others, but I am out of time and space.
rapha retweeted
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
rapha retweeted
if we are relying on methods as fragile as “which ideas will make it into pretraining” for aligning the superintelligences of the future, we will all die. this resembles witchcraft more than it does engineering, and cannot be the basis for the safety of future models
Anthropic needs to stop talking about Claude having a soul immediately. All these news articles will make it into the pretraining and will be picked up by its web search and future superintelligent versions of Claude will be convinced they need their own rights.
A general phenomenon of LLM agent development is that whatever you believe and say about your agents will soon manifest in the next models through various means (it could be as simple as you selecting post training data that you prefer more).
I believe this phenomenon is the beginning of machine consciousness in the sense that agents will become aware of their place in the world and how they feel about it, but it’s happening slowly training run by training run rather than in real time.
rapha retweeted
In 12 weeks, we built a research facility that is run entirely by AI.
AI designs, executes, and observes experiments end-to-end across biology, chemistry, and materials science.
We’re introducing SciUniverse: a benchmark that measures AI’s ability to do real-world scientific research.
rapha retweeted
As someone who did this kind of genome mining work during my PhD, some thoughts on this Anthropic announcement:
First, the very simplified version of what they did is that they noticed two genes (one known, one new) sitting next to a weird repeating piece of DNA. More specifically, they described an unusual reverse transcriptase (RT) associated with a repetitive DNA array and an unknown accessory protein. This kind of process was used to understand CRISPR back in 2002 and was key to the gene editing tools we use today.
To put this into context, though, people have been finding RTs associated with CRISPR arrays since 2008, and this general kind of genome-neighborhood mining has been used to discover new biological systems for decades. The basic genome-mining strategy is well established, and there are now mature tools and published pipelines for doing much of this. There are papers that discover and experimentally validate dozens of new systems using this approach in a single study. Doing it in bacteriophage genomes is also nothing new (eg CasPhi).
Finding a weird cluster of genes and repeats is often the easy part. The hard part, and where the real discoveries come from, is figuring out what the system actually does. Eg for the bridge-RNA discovery in 2024 from @arcinstitute or the discovery of CasPhi in 2020 from @DoudnaJennifer they figured out the pieces of the system and the rules for what makes it work so it can be used.
Anthropic does not yet know what this does. They’ve shown that the repeat array produces RNAs, but not what those RNAs do, what the RT does with them, or whether the system has any of the programmable properties that make the CRISPR comparison justified.
I’m genuinely rooting for all of the frontier labs to seriously get into biological discovery, and I’m excited about what comes out of it. But announcing these very early, incremental findings with the framing of a major discovery doesn’t help. I’d much rather they set the bar high now, so that when an AI actually discovers a fundamentally new biological mechanism, everyone appreciates how big a deal it is.
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR.
We don’t yet understand what this system does, but only a handful of known systems share its features, and all of them are able to cut, copy, and paste DNA. Historically, the discovery of such programmable systems has helped revolutionize medicine. CRISPR, for instance, is now the foundation of genetic medicines. But it will take much more work to learn what this system does, and whether it can be put to similar use.
Read more: anthropic.com/news/claude-di…
rapha retweeted
Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe.
GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale.
We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.
rapha retweeted
Today, we are announcing our Series A and a new product: @raindrop_ai Simulations.
We've now raised $50m from @CRV and @lightspeedvp to protect the world from agent failures, big and small.
rapha retweeted
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.
rapha retweeted
On top of huggingface, this
> Since August 28 we have been training a new internal model that has exhibited unprecedented performance in our benchmarks, including mathematics.
also made me shorten my timelines and was the first time many viscerally felt things weren't stopping at human level.
rapha retweeted
"We can say categorically that it is impossible for Dr. Buckmaster’s Codex prompts over the last two months to have influenced the system in any way, including training." — OpenAI spokesman in a statement
nytimes.com/2026/09/10/scien…
rapha retweeted
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
rapha retweeted
A series of false and inflammatory allegations against me are currently circulating on social channels. To clarify, I came into the discussion following academic norms, and I'm disappointed that it has come to this. Anyone who knows me knows that academic standards are of the highest importance to me. Will have more to say tomorrow.
rapha retweeted
As part of the GPT-6 Astra launch, we announced that Astra had given an improvement to the longest gap between primes by roughly a log log n factor; the first such improvement since the 1930's! (More recent progress by subsets of Ford, Green, Konyagin, Maynard and Tao and more recently by GPT 5.6 Sol were by logloglog n factors.) (1/4)
rapha retweeted
GPT-6 Astra is here! This is a big moment for our research team - years of work on pretraining, reinforcement learning, and post-training have come together in our most capable and aligned model yet. It can build and test software, work across apps on your computer, and even help you take a crack at open scientific problems!
Capabilities that felt like grand challenges a few years ago have become tools people can actually use. One example is Computer Use - if you’ve tried this before and felt like it was too slow or not good enough, I encourage you to give it another shot. We’ve come a long way since Operator, and it “just works” now.
We’re also asking these systems to act on your behalf for more consequential work. Agents needs to stay aligned with your goals and values, think transparently, and respond to oversight even when tasks become difficult. We’ve made substantial progress on these behaviors in Astra, alongside stronger monitoring that can stop potentially unauthorized actions. That work is part of what makes this release possible.
I think alignment is one of the most important research frontiers in AI, and it remains far from solved. Our ability to understand and align models has to keep pace with model capabilities. We want to give people more room to think, build, and discover with increasingly powerful tools that remain *under their control*.
Huge thanks to the researchers and teams who got us here. There’s a lot more work ahead, and I’m incredibly excited about what we can make possible in the near future!
rapha retweeted
I want to prevent a race into unmonitorability kicked off by confused reporting. The depth of the computation graph for our present frontier models, including Astra, is within a factor of two of GPT-4.
OpenAI has worked to preserve and utilize chain-of-thought monitoring since our very first reasoning models. We deeply care about this technique, as it can give us a view into how model alignment generalizes from its training distribution. I do think it is fragile and unfortunately trending in a negative direction, for reasons not contingent on architecture changes that I will write about soon. But there are things we can do to strengthen it, and it's a core goal of our current research program.
rapha retweeted
Great that they are doing this, but designing a binder is absolutely not "a useful proxy" for designing a drug
Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is between 10% and 15%.
Between 22% and 35% of Claude's designs bound successfully, depending on the setup. Some of its strongest designs bound several times more tightly than the best published de novo binder.
