winning without trying
Joined March 2023
- Tweets734
- Following12
- Followers9
- Likes12.8K
void retweeted
The speed is astonishing
A globular springtail appears to teleport
it launches itself using its spring-like furcula in a split second.
void retweeted
Küçük et parçasıyla kandırılmaya çalışan yavru kedi, kendisinden daha ağır bir eti kapıp kaçtığı anlarla viral oldu:
void retweeted
drug discovery has a dirty secret: for most proteins linked to disease, we have no idea what they actually do. because we have no way to turn them off and watch what happens. that's what "tool compounds" are for. not drugs; switches.
Vanderbilt just published the first selective switch for TAOK-1, a protein tied to alzheimer's that's been basically unstudied for exactly this reason. they built a whole library of related molecules, tweaked atom by atom, until one (VU6083859) hit TAOK-1 and nothing else.
then, the universe smiled: a different molecule in the library (VU6080195) turned out to ACTIVATE the entire TAOK family. the opposite of what they were looking for.
now the field gets to ask two questions instead of zero.
people think drug discovery is finding drugs. most of it is building the lamps under which the drugs can be found at all.
void retweeted
i have something to say. i live my life very intensely. i need melatonin to sleep because i have insomnia. because i know i am holding a technology in my hands that can uplift people, and every second wasted is a life lost.
this is not a sob story, it's a cool one: i was recently prescribed low-dose anti-anxiety meds, which is quite ironic for someone trying to change the world. they have helped me become more productive. but they have not made me any less ambitious; if anything, more ambitious. i sleep, i wake up, and i work. but my work is more like a mission.
a way in which i am connecting with the entire world. when you plug into a socket, you're connecting with a power grid. when i wake up, i feel like i am connecting with all of humanity. i know that sounds extremely cringe, but it is so true.
i understand why people cannot yet grasp exactly how many lives precigenetics will save. i understand that people are skeptical, but that is a waste of time to me. because science is science and science says that our approach works.
i wear multiple hats, and i must execute so darn well. i do my best, every single day. but slowly, i am letting myself realize that i cannot fight my purpose with anxiety. i can only flow into it.
i love human beings, and i hate tragedy. i understand that not everyone understands just how huge this is, and exactly why it works.
at some point, i realized, explaining it to literally every single person is taking me away from the singular thing that will actually matter - me just doing it. i feel myself drifting away from the status quo and the Overton window into a world that will, but does not yet, exist in every person's consciousness.
but it is not lonely. it is weirdly exhilarating. it brings me energy to build something so crucial, that every part of a normal life, a normal job, i thought i would have growing up, has evaporated.
nothing about my life is normal, and i say that in the best way possible. perhaps the coolest part, though, is that this is extending beyond me, slowly but surely.
it is becoming a responsibility the team i have built takes on with great passion. it is getting easier and harder at the same time.
i find myself spending far less time consuming and far more time creating. it is like being a scuba diver that just started the descend, holding their ears shut, easing into the pressure 100 feet below earth's surface. but one who will inevitably come back home, and when they do, they'll bring with them cures, and gold, things that don't just change their life, because that is not half as fulfilling. they change the entire world.
and, it just so happens, i am a quadruple-certified scuba diver. so i think i got this.
void retweeted
My cancer case is n=1. I'm not holding it up as a model, because most of what I did is not specifically replicable or scaleable. But it is a proof of concept that fully personalized medicine, built around one patient's own biology, is viable in ways it simply wasn't a few years ago. I have lived a version of the future of cancer care that I think is worth aspiring to more broadly.
Many of the tools that made this future possible for me are getting better, faster, cheaper, and more accessible every year. I think what we need next is the infrastructure to deploy these tools, and the community willing to build this infrastructure, one patient at a time.
I want to help that future arrive faster. I wrote an article in The Cancer Letter cancerletter.com/guest-edito… for the physicians and researchers closest to the patients who could benefit most from this approach, who also believe in this future. If you are working towards building the systems and practices that will make personalized medicine operationally real, we want to collaborate with you too.
void retweeted
PSA: Most biglab people now read almost zero papers and understand ICLR/ICML/NeurIPS to be mainly full of overclaims & fraud. (but there are a few diamonds in the rough of course)
void retweeted
Thanks for posting this thought-provoking article. Readers should read it with an eye towards the three-level ladder of causal understanding -- parallels may pop-up and, when they do, we have the beginning of mathematizing intuition.
Google Deepmind argues that LLMs can never make real scientific discoveries.
They published a paper breaking down Albert Einstein’s private view of scientific discovery.
In a famous letter to his friend Maurice Solovine, Einstein drew a diagram of how science actually happens.
It is a cyclical loop.
First, you experience raw sensory data. Then, through a mysterious, non-logical act of intuition, you make an intuitive "jump" to abstract axioms. Finally, you use strict logical deduction to derive consequences from those axioms.
Generative AI has completely mastered two-thirds of this loop.
• Induction: Statistical pattern matching across billions of tokens.
• Deduction: Formal proof generation, like AlphaProof solving complex math Olympiads.
AI can crunch data and it can prove theorems.
But it cannot make the jump.
The paper argues that AI completely lacks Abduction, the generation of novel explanatory hypotheses when observational data is scarce.
The prevailing tech myth says that "creativity is just data compression." That if you feed an LLM enough text, scientific breakthroughs will naturally pop out.
Einstein’s formulation of General Relativity proves that is a delusion.
When Einstein formulated relativity, the observational data didn't demand a new physics framework; classical mechanics was still massively successful. The breakthrough required a conceptual rupture. An intuitive leap from physical reality to a brand-new set of foundational axioms.
An LLM can execute the math once the axioms are given. But it is structurally incapable of formulating those premises on its own.
It can interpolate inside existing human thought, but it cannot transcend it.
The translation of physical reality into formal axioms remains the absolute, hard bottleneck of artificial scientific invention.
We can build models with trillions of parameters. We can scale compute into the stratosphere.
We can make the calculator infinitely fast.
But until we solve grounding, the machine can process all the data in the universe.
It still can't make the jump.
void retweeted
Another innovation which I've not seen in frontier models is Attention Residuals. They make an astute observation that what made attention effective in the first place beyond standard RNNs is the ability to access state from previous steps, instead of summarized in the state from only the previous step.
Standard Transformers try to get around this depth-wise with residuals, x_t = x_{t-1} + f(x_{t-1}), which if you notice basically looks like an RNN. Since attention helps RNNs, it should help Transformers --> Attention Residuals. Great idea, but to actually get this to work efficiently at frontier-scale is impressive. Again, high-entropy move.
void retweeted
Our new paper brilliantly lead by @itamarlandau "A predictive theory of experimental design for inferring neural population geometry in large-scale recordings." biorxiv.org/content/10.64898…
Our theory can say a lot about the geometry of new data *before* it is collected, by extrapolating from past data. We quantitatively predict:
1) How neural dimensionality, and the reliability of neural correlations and individual neural PCA modes grows with neurons and trials. This allows one to design experiments before the data is collected.
2) We find a "blessing of dimensionality" whereby recording more neurons actually allows us to record *fewer* trials while still reliably inferring neural population geometry. This opens the door to new types of experiments with many more complex trial types.
3) We develop scaling laws for neural prediction using masked autoencoders - a key technology for building foundation models in neuroscience. In the simple setting of linear autoencoders on a single session, we find power law behavior of prediction performance and autoencoder size with neurons and trials, and we trace these power law exponents for neural prediction to power law signals in collective neural modes of the brain.
4) We test our theory across multiple species (mouse, monkey, human) and recoding modalities (electrophysiology, calcium imaging, and fMRI).
See @itamarlandau's excellent thread for more information: xitter.cf/itamarlandau/status/20…
Also, yet another fun collaboration with Mark Schnitzer!
Neuro experiments capture 10,000+ neurons but often with only a ~100s of trials. So the data is ~100s points in a 10,000-D space. Can we trust the extracted "population geometry"?
This motivates our new paper: a predictive theory of experimental design.
@SuryaGanguli
1/13
void retweeted
Which is much harder than just getting rewards every time you get diamonds or greens.
The authors recorded brain activity throughout the entire time while the monkey learned this task.
void retweeted
Introducing "Diffusing Blame": can a neural network learn competitively while strictly obeying Dale's principle, the rule that real neurons follow? We show it can, across both image classification and reinforcement learning. 🧠
Accepted at #ALIFE2026
arxiv.org/abs/2606.31700
Real neurons generally follow Dale’s principle: each neuron is predominantly excitatory or inhibitory. Standard artificial networks usually ignore this constraint, allowing every unit to mix positive and negative outgoing weights.
Backprop makes the gap even wider. Its backward pass needs exact transposed copies of the forward weights, the so-called "weight transport problem,” which biology doesn’t seem to have a mechanism for.
So we asked: can a network that strictly enforces Dale's principle still learn well, without weight transport?
Our approach builds on Error Diffusion (ED), a local rule that routes a single global error signal directly to every hidden unit, where each layer is split into separate excitatory and inhibitory streams with four non-negative weight matrices, so a synapse's sign comes from fixed population identity rather than a learnable weight. Our main contribution is to extend ED from binary to multi-class problems via modulo error routing.
We then asked whether this routing mechanism could provide useful credit signals in the noisy setting of reinforcement learning. During PPO training on Ant, Humanoid, and HalfCheetah, we compared each local ED update with the corresponding true backpropagation gradient. Among the routing schemes we tested, modulo routing consistently produced the strongest alignment.
Taken together, these results show that Dale-constrained networks can still learn without transporting weights backward, suggesting a potential path toward learning rules that are both effective and more biologically plausible.