Why Did a Larger AI Get Better at Surviving Broken Arithmetic?
Researchers found that language models trained amid simulated arithmetic faults grew more resilient, but proof on real chips is still missing.
Puzzle
Why Did a Robot Need to See the Mistake, Not Just the Move?
A smooth robot demonstration can hide the crucial correction. New research finds why showing recovery may teach machines to align better.
Why One More Step Can Shrink a Network’s Required Seed Set
A directed-graph puzzle reveals why one more step can make a tiny set reach every node—and why the two-step version remains open.
Why Did Better Instructions Still Fail on the Tiny Chip?
On tiny microcontrollers, compiling code is not proof it will work. Real measurements gave optimization agents clues static instructions could not.
The Robot Waited Five Seconds—and Mistook Waiting for Work
Harmless pauses made a robot stop counting early, exposing a memory shortcut that confused elapsed time with real progress.
Could a Carbon Planet Pass for an Ordinary Rocky World?
New models trace carbon planets from graphite to diamond—and reveal why their size may still be hard to distinguish from ordinary rocky worlds.
How a Random Ordering Turns Average Degree Into a Guaranteed Tree
A six-decade graph theory puzzle gets an elegant answer: random vertex orderings show when average density must contain every tree.
How a Quantum Simulation Can Catch the Heating That Warps It
A trapped-ion simulation may use Gauss’s law to spot heating-damaged runs, but missed event pairs and unequal rates leave a residue.
Could a Dark-Matter Detector Hear the Sun Through Its Own Noise?
A dark-matter detector separates solar neutrinos from radioactive noise, opening a new low-energy window while exposing the limits of the signal.
When Every Answer Is Right, Why Do Humans and AI Choose Differently?
In 270 multi-answer puzzles, humans spread choices more widely than AI models, revealing what accuracy scores miss—and what they can’t explain.
How a Fading Charge Stayed Free in a One-Dimensional World
A 50-year-old puzzle in a one-dimensional quantum world reveals why a charge can fade toward zero without triggering confinement.
The AI’s Doubt Rose for Vague Clues—Then Failed on Look-Alikes
A collaborative puzzle reveals why AI confidence can miss look-alike objects—and why a poorly targeted warning may mislead its human partner.
Why Did a Robot Solve a Lockbox Better With Worse Data?
A robot solved a mechanical puzzle more efficiently with ordinary images than perfect state data, exposing a surprising link between noise and reasoning.
A Pulsar’s Companion Changes Its Light—and Defies the Usual Models
A confirmed millisecond pulsar has a companion whose changing light curve resists simple explanations and complicates its neutron-star mass estimate.
Why Did Four Parallel Photonic Hinges Send Waves One Way?
A 3D microwave crystal sends topological waves the same way along parallel hinges, surviving tests with a corner and missing rod.
A Stellar Stream Bends the Wrong Way—and Dark Matter Takes Notice
A new geometric test could use the curves of shredded star streams to tell whether dark matter halos are round or triaxial.
The Galaxy Survey Stranger That Turned Out to Be a Planet-Mass Drifter
A red speck in JWST’s deep galaxy survey proved to be a nearby, planet-mass Y dwarf—with an atmosphere current models struggle to explain.
Why Does a Young Binary’s Hot Gas Retreat Between Accretion Bursts?
JWST and ground-based spectra reveal why DQ Tau’s hot molecular gas may retreat between stellar feeding bursts—and what remains unexplained.
When a Space Robot Loses Its Thrusters, Can It Still Arrive?
A space robot might reach its target even after losing every thruster—if its route is planned around the right moment to coast.
Is This Black Hole Finally Ending Its 34-Year Outburst?
A decades-long black-hole outburst is fading, but thick gas hides its X-rays. Radio clues suggest quiescence—not a settled verdict.