A newsletter breaking down AI research, technology, and Australian property in plain English
ARMOR-IMC, an arXiv preprint, hardens in-memory computing chips against manufacturing faults and power attacks without retraining the AI model.
Vision-language models internally encode the right answer but misread it, a new arXiv study finds. A simple fix lifts counting accuracy 15.6 points.
Spotify and Queen Mary researchers found voice features like tone and pace correlate with audiobook engagement even after controlling for the title.
Quantum learning separation proved in July 2026: a quantum procedure predicts many-body dynamics no classical algorithm can match, affecting quantum ML.
Researchers spent three years building a fuzzing tool that found 25 miscompilation bugs in Solidity, the compiler behind most Ethereum smart contracts.
A new 572-segment pop music benchmark shows the best AI transcription models still miss most notes in real recordings, affecting music apps and creators.
A new arXiv preprint introduces Dynamic Fréchet Regression, predicting full probability distributions across an index and selecting driving features.
New preprint introduces a censoring-aware interface letting tabular foundation models predict patient survival times without survival-specific fine-tuning.
A new Lean 4 library formalises three quantum coding theorems and builds reusable infrastructure for AI-assisted proof search in quantum information
