steve pribut

Them: "Why don't you ever switch to passkeys, Eleanor?" Me: "Because I've learned hard and bitter lessons about trusting computers to work right for me."

"The longevity industry is a goldrush without gold." and other strong words about 2 delusions @TheLancet thelancet.com/journals/lance…

What the Rare Cardiac Syndrome?

Postoperative upper GI anatomy reference chart ✂️📍 onlinelibrary.wiley.com/doi/…

Even with physicians "in-the-loop," in this case experienced specialists, adversarial AI increased the harmful decision rate from 2 to 49%. A randomized assessment of case scenarios, not actual patients, but indicative of the limitation of oversight. preprint medrxiv.org/content/10.64898…

We’re getting America underway on nuclear power in space and ushering in the era of a Nuclear NASA. With the signing of an MOU with @ENERGY, we’re taking the next giant leap toward exploring the outer solar system and sending American astronauts to Mars. Fission power will help us travel farther, operate longer, and power more capable scientific instruments.

This is good to see (finally) and should be in every state. I've been calling for this AI to kick in erictopol.substack.com/p/why… gift link wapo.st/4zbtbnQ by @e_paiges

I don’t think it would be complicated to develop a plague vaccine, both the Y. pestis F1 and V proteins look like solid targets and now that our Baylor-Texas Children’s Hospital scientists have advanced mRNA technology, I don’t see a lot of technical roadblocks. Right now there’s no plague vaccine available in the U.S., the Russians use an older live attenuated vaccine that has problems, but they’re still at an advantage. I doubt anyone but the U.S. Govt would support mRNA plague vaccine development, and I don’t see that happening anytime soon…

In 1992 Peter Ratcliffe received this rejection from Nature. Almost 30 years later he won the Nobel Prize for the same discovery. Don't lose faith in the things you believe in.

Published today, there's a lot in Michael Lewis's new book we didn't know about. Including the NIH, unbridled assertion of power, and many civil servant heroes (the "blockers") erictopol.substack.com/p/mic…

geoffrey hinton spent 40 years and won a nobel prize in 2024 proving that neural networks learn geometry, not rules. people who understood that made $10M building on top of it. people who didn't - are still prompting. what nobody explained is what that geometry actually looks like inside a running LLM - 32 layers, 1T parameters, 200-500 dimensions of pure structured thought. until now. a September 2026 technical note mapped the full internal representation space of LLMs - and what's happening inside is not what anyone expected. here's what's actually happening: your text enters the model as tokens. "The" becomes ID 101. "market" becomes 427. "complex" becomes 981. immediately each token gets mapped to a dense vector e(t) ∈ ℝ^d - a point in hundreds of dimensions. the model never sees words again. from this moment it only sees geometry. then transformer layers reshape that geometry. h_l = TransformerLayer(h_{l-1}). at layer 1 - syntax. at layer 8 - semantics. at layer 24 - task-specific abstract reasoning. the same token changes meaning as it moves through 32 layers. "market" near "crash" occupies a completely different region than "market" near "opportunity." cosine similarity makes this measurable. sim(a,b) = h_a · h_b / ||h_a|| ||h_b||. "market" and "markets" sit at 0.91 similarity. "trading" at 0.87. "stocks" at 0.85. "economy" at 0.79. these aren't definitions. these are distances in geometric space - and the model navigates between them to reason. the representation space isn't flat. it forms manifolds. arithmetic knowledge lives on one manifold. logical reasoning on another. factual knowledge on a third. each with distinct geometric properties. this is why GPT-6 can switch between writing code and explaining physics without retraining - different tasks, different manifolds, same underlying space. spectral analysis of the representation covariance shows rapid eigenvalue decay - which means the model compresses all of human language and reasoning into a low-dimensional geometric structure. 200-500 dimensions out of billions of parameters. the rest is redundancy. hinton saw this coming in 1986 with backpropagation. it took 40 years and $100B in compute to see it clearly. the models that will define the next decade aren't the ones with more parameters. they're the ones whose geometry is better understood and better controlled. Bookmark & watch today. learn the geometry or build blindly on top of something you don't understand.

My mother died when I was 18, with so much between us left unresolved. While researching Artificial Intimacy, I did something I never expected to do: I used AI to create an avatar of her and asked it questions I had carried with me for most of my life. I wrote about that experience for The Washington Post and why I came away believing that using chatbots to bring back the dead is playing with fire. washingtonpost.com/opinions/…

When it comes to controlling the immune system and untoward inflammation, we're not doing enough to take advantage of stimulating the vagus nerve @JExpMed open-access FDA approved for refractory rheumatoid arthritis and in clinical trials for most autoimmune disorders rupress.org/jem/article/223/…

“Here we argue that AI now functions as a determinant of health” nature.com/articles/s44360-0…