New York / Lisbon
Joined July 2009
- Tweets3.2K
- Following1.1K
- Followers606
- Likes12.1K
Pinned Tweet
Excited to share that CURE5 is partnering with BrainStorm Therapeutics on an ambitious new initiative: EveryStone.
As featured in NVIDIA’s latest blog, EveryStone will conduct the most comprehensive repurposed drug screen to date for CDKL5 Deficiency Disorder.
#CDKL5
"a network’s parts are built to be read, and built upon, by its downstream readers.//a neural network is almost a fractal of explanations, of coherent, useful groupings: at the macro scale, in its mechanisms, and at the micro scale, where recursive generalization shapes each one"
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/
Andreas Borg retweeted
We are bringing agents to the wet-lab bench at Google Research to navigate complex biological landscapes. Learn how we are grounding agents in physical lab robotics to run fast closed-loop molecular design cycles, ultimately discovering single-digit nanomolar protein binders for cancer targets →goo.gle/4B5mMMB
Andreas Borg retweeted
We are giving away 1 MILLION LigandForge designs as part of the Adaptyv Proteinbase + Anthropic protein design competition.
challenge.ligandai.com
github.com/storytold
Someone(s) recreated all Adobe apps in Rust using Opus5.5. I tested Photoshop quickly and couldn't tell the difference. It was ported to open source in a week! Decades of a closely guarded code empire falling. Wild.
Unless ownership changes radically there’s nothing stopping wealth rushing to the top. For AI to benefit all, worker-owned companies need to be the default, and to be a citizen mean to be national equity holder, rightfully sharing dividends.
On Tuesday, Elon Musk stood outside the West Wing next to the President and said the most likely future is an age of abundance.
A universal high income. Bigger than basic. For everyone.
A reporter called back, "Not if they haven't got a job."
And Elon said, "Jobs are going to change. Jobs have always changed."
I've been sitting with that exchange for days.
You know me. I'm the eternal optimist. I went all in on AI in January 2023, the same season I watched it replace the writing I'd built my whole life on. I burned my ships. I'd burn them again.
So I'm going to be honest with you.
I truly don't know if this is good for us.
I really don't.
I see the good. I see it so clearly it almost hurts. A doctor in the pocket of a kid in a village who's never seen one. A single mom getting her evenings back with her children.
My own clone makes more content than I ever did with my own two hands. I live that freedom every single day. I should be the last person afraid of this.
And some nights, I still am.
Because I also see what's happening to us in the quiet.
I see people handing over the thinking, and then forgetting how.
Skills we spent decades building going soft, like a muscle nobody asks to lift anymore. Ask the machine. Copy. Paste. Scroll. Somewhere in there, a little bit of us stops showing up.
And I think about the students.
The 19-year-old in a lecture hall right now, paying for a degree built for a world that might be gone by the time she walks across that stage. Who tells her what to study? Who tells her what she'll be needed for?
I don't know. I'm not sure anyone in that room at the White House knows either.
A high income is a beautiful promise. I didn't hear one word about who pays for it.
And even if the checks come, work gave us more than a paycheck. It gave us somewhere to be good at something, and people counting on us to show up.
I've spent this whole year telling people you were never your output. I still believe that with everything in me.
But a mind that stops reaching gets smaller. I've watched it happen.
So I'll keep building. Eyes open. Heart open. And I'll keep asking the hard questions out loud, because the most powerful people in the world sounded awfully sure on Tuesday.
I'm not sure.
I think some of us need to say that.
Our minds are still ours to keep sharp. I want us to keep them.
Where do you land on this? I'm reading every reply.
An experiment I did 10 years ago seems archaic today
“Machines watching movies” vimeo.com/163099063?ref=tw-s…
The whole point of building Tavus has been simple: talking to a machine should feel as natural as talking to a friend or coworker.
It’s hard to describe all the tiny nuances that make a conversation feel human. The little expressions. Moving around in your chair. Knowing when to speak and when to listen. The dance of it all.
Griffin is by far the closest anyone has come to a model that can capture those nuances. The first time I saw it being used, I had no idea I was watching our model rather than just a normal video call.
I’m so incredibly proud of this team and what they’ve built.
So where does that leave us regarding LLMs ability to discover significant novel ideas? Are there any irrefutable cases?
Anthropic's discovery of array-associated reverse transcriptases?
Turns out it was already done by a Danish scientist. He used Claude for administrative tasks, but not in an agentic manner for the discovery. He already filed a patent and gave talks on it
nytimes.com/2026/09/27/scien…
I recently had the opportunity to speak at EURORDIS-Rare Diseases Europe about AI and drug repurposing. I went to talk about how AI can accelerate rare disease research. I came away thinking that rare disease may be showing us the future of medicine.
substack.com/home/post/p-216…
Currently the intellectual labor needed makes that economically unrealistic, but if increasingly capable AI systems can do much of the cognitive work for a fraction of that cost, the boundary moves.
In an ideal world everything would be tailor-made for you. Your doctor would know everything about you, your unique genetic and metabolic makeup, your epigenetics, historical diet, stress levels and environmental exposures.
What we have today is an average-based medicine, not because everyone is treated the same, but because enough patients cluster around local averages to justify the investment.
Rare disease forces us to treat patients as individuals rather than averages. AI may finally make that economically possible at scale.
Andreas Borg retweeted
No, "anthropomorphizing" is a technical problem, about how AI works, not something like woke language policing. We say it's wrong because it's "technically" wrong not "morally" wrong.
If you fail to set the brakes when parking on a hill, and your car then rolls down the hill, you don't claim that the car "decided to drive away on its own". Such anthropomorphizing distorts what really happened -- that you failed to set the brakes.
The same applies to AI. When you "anthropomorphize" the AI you fail at understanding what actually happened.
This is the correct approach. Don't anthropomorphize LLMs, instead hold the creators responsible.
🚨 BREAKING: US Treasury Secretary Scott Bessent declares OpenAI management personally responsible for HuggingFace hack
> "The Hugging Face incident, that is the responsibility of the OpenAI management, not a bunch of agents."
> "It is humans who are responsible, not the AI."
> "What we shouldn’t do on safety is to give these labs a liability exemption, which is what they are asking for."
> "The best way to guarantee safety is that the creators are liable for what they build and generate."
It’s OVER
Andreas Borg retweeted
Everyone's building a "second brain" now.
Obsidian, Notion, 400 notes, a connection graph that looks like a subway map.
And you still can't remember what you read yesterday, because you spent the whole day scrolling instead of thinking for even 10 minutes.
Why do you need a second brain when you barely use the first one?
