Humanoid safety filters failed in selected physical tests
Seven selected filter trials produced collisions; broader search totals conflict.
Explore by topic
Science & Health
Explore BIG CHANGE reporting on Science & Health: what changed, how it works and why it matters.
Analysis
47 stories
PaperFormalizer asks authors what Lean should prove
PaperFormalizer pairs author review with Lean proofs of revised claims.
DiffDreamer gate physics
Simulated gates, externally tracked flights.
Skin-lesion test separates diagnosis accuracy from discharge risk
CNN led diagnosis; its preset discharge rule missed more melanomas.
SpecEvo reserves costly model calls for program-search interventions
SpecEvo tests sparse frontier calls in program search.
Real PubMed IDs can still point to the wrong evidence
A PubMed ID identifies a citation, not evidence for a claim. A biomedical agent study shows why a resolving link still needs a check of the cited abstract and verdict.
Language-guided proton planning and its limits
How a research prototype turns instructions into planning adjustments.
LitPivot ties proposal revisions to paper rankings
LitPivot connects proposal edits with literature.
IORA separates geometry from holding strength
Robot grasp planners need to check both contact geometry and the forces a hand can actually produce.
Eko test: notes aided tower building
Eko note aided towers.
Agent4Sci: evaluating engineering gains and scientific insight
A power-flow intervention clarifies Agent4Sci’s evaluation boundary.
GelSlim 5.0 adds vision to touch
A shared viewpoint connects object appearance and contact.
NVIDIA maps Microduck obstacle training from AI agents to simulation
NVIDIA details an agent-to-simulation Microduck workflow and an untested physical handoff.
Study traces dense AI prose to measurable language patterns
An October preprint finds several processing difficulty proxies more often in some model replies than in comparison text. A separate experiment shows that a length penalty can shift one small model's grammar, with limited transfer.
Google's ForTy forest map goes beyond tree cover, with limits
Google's ForTy map separates natural, planted and farmed tree cover. Its 2020 baseline and early annual series offer new questions for forest monitoring, with significant class-level uncertainty.
EgoExplorer records 100 hours of human work for robotics research
A wearable system pairs video, trackers, gloves and touch across 100 hours of human activity. Its full-body motion is reconstructed, and robot replay remains a two-clip simulation.
Community hub organizes revisions to OpenAI’s math papers
A community fork of OpenAI’s mathematics collection is recording manuscript edits, reviews and follow-up work. Its growing notes do not certify the proofs.
Continues From unit distances to Navier-Stokes: AI mathematics reaches a new phase
Claude-designed proteins bound four scorpion toxins in lab tests
Anthropic reports that 17 of 211 selected protein designs met its toxin-binding criteria. The study did not test whether they neutralise venom in cells or animals.