Backpropagation on digital chips just got a pink slip - or at least, a memo suggesting it start updating its resume. A new paper from Lawrence Berkeley National Laboratory shows that physical systems running on nothing but ambient heat can be trained to do machine learning, using the same gradient...
Two types of people exist in the world of environmental chemistry: those who already know that predicting how long a chemical lingers in soil is a nightmare, and those who are about to find out.
Here's the lay of the sea. We have two approved malaria vaccines now - RTS,S (Mosquirix) and R21/Matrix-M - and they're saving lives across Africa. But they're like boats that can only handle calm waters. RTS,S clocks in around 39% efficacy over four years (Laurens, 2020). R21 does better, hitting...
We find ourselves, dear reader, in the grip of a delightful pharmacological contradiction. Immune checkpoint inhibitors (ICIs) - the crown jewels of modern cancer therapy - work by unleashing your immune system to hunt down tumors. Splendid idea, truly. Except the immune system, once unchained,...
Ant colonies don't have architects. No single ant draws up blueprints for the tunnel system - they just try stuff, keep what works, and let the colony self-organize into something weirdly optimal. Cao et al. looked at the absolute mess that is hyperspectral unmixing network design and said: what if...
A routine Tuesday in a computational biology lab: someone feeds a neural network millions of fake cells dividing, and the network quietly figures out something that decades of equations couldn't solve.
Like a colony of leaf-cutter ants, each hauling a tiny fragment back to the nest to feed the fungus that actually nourishes the whole operation, AI tools in medical research have quietly organized themselves into a division of labor that most researchers are still figuring out how to manage.
This is a paper about two proteins, SOX9 and YAP1, that keep bile duct cancer alive by covering for each other whenever doctors try to knock one of them out.
Here's a question that should bother you: what if the billions we're spending to clean up contaminated soil is based on numbers that don't mean what we think they mean?
Designing anion exchange membranes used to be like renovating a house by randomly ripping out walls and hoping the roof doesn't cave in - the old approach was slow, empirical, and occasionally catastrophic. SPARK is the structural engineer who finally showed up with actual blueprints.
In 1986, the FDA approved a blood test for a protein called prostate-specific antigen - PSA - and men's health screenings were never quite the same. Not in the good way. For four decades, doctors have been squinting at PSA numbers like they're reading tea leaves, because here's the dirty secret of...
Ladies and gentlemen, it is the year 2030. Every weather radar station on the planet runs a neural network so lean it fits on hardware your grandmother's microwave oven would be embarrassed by - and your seven-day forecast has never been more accurate. The ghost echoes, the phantom storms, the...
A search-and-rescue drone cuts through morning fog over collapsed rubble. Its camera scans for survivors - tiny figures against a chaotic landscape. Somewhere in its slim chassis, a neural network fires. Not the heavy, power-hungry kind that would drain the battery in twenty minutes, but something...
Ever try to make caramel and have it go from "almost there" to "burned disaster" in about two seconds flat? That's basically what chemists deal with when making polyurea coatings through interfacial polymerization - the reaction between their two ingredients happens so fast that controlling the...
That sounds like the plot of a mediocre sci-fi movie, but a team from Chalmers University of Technology and Tsinghua University just did exactly that. Their new model, Human2, is a genome-scale metabolic model (GEM) of the entire human body - and they used large language models to help build it....
Good news: someone figured out how to make time series foundation models actually work across wildly different datasets by changing just one line of code. Bad news: the reason they had to do this is that our best foundation models have been quietly falling apart whenever the data looks even...
Can a neural network run on hardware where a single "bit" is literally in two states at the same time - and if so, does that actually help?
The evidence was right there in the joint. Not a human joint - a robot joint. Specifically, a flexible one. See, most robotics textbooks pretend that when a motor turns, the link it's attached to moves in perfect lockstep. Neat. Tidy. A beautiful lie.
A storm has been brewing in meta-learning. The whole promise of "learning to learn" - training AI systems that can pick up new skills from just a handful of examples - runs into a brutal reality check the moment your training tasks are noisy, incomplete, or just plain bad. Jun Shu and colleagues...
As of early 2025, the best anyone could do to classify a rare pediatric sarcoma was ship tissue slides across the country to one of a handful of specialist pathologists, wait weeks for results, and maybe order a $5,000 genetic test on top. This paper changes that.