Technology & AI
ALER-TI fills time series gaps using historical retrieval
ALER-TI, a new arXiv preprint, retrieves cached historical patterns to reconstruct missing time series values, tested across six real-world datasets under varyi
Not A Tech Guy
AI and technology, explained through what they actually mean. An AI-assisted newsroom under human editorial rules — every story cites its primary sources so you can check them yourself.
127 stories
Technology & AI
CNeVA traffic sim adds controllability top models lack
CNeVA lets developers steer virtual driver aggression and caution on the Waymo benchmark, offering per-channel control that higher-ranked imitation models lack.
Technology & AI
ActionCache speeds robot AI 34× without retraining
ActionCache cuts inference latency for vision-language-action robot models by up to 34× without retraining, a potential breakthrough for real-time robotic deplo
Technology & AI
G-RRM neural solver hits 33× speedup on Sudoku — with a catch
G-RRM uses neural reasoning models to guide classical SAT solvers, cutting Sudoku backtracking 33× — but the speedup vanishes if the solver can't reject bad hin
Technology & AI
6,588 GitHub repos tagged with NAICS industry codes
NAICS-GH labels 6,588 GitHub repositories with industry sectors at 96.98% precision, enabling research into where open-source code actually comes from.
Technology & AI
LLM refusal neurons: causal audit finds no unique mechanism
A preprint audit across five LLMs finds refusal lives in a redundant subspace — and trusted rank-stable methods can be the least causally valid.
Technology & AI
MedPMC: 11M clean medical images beat bigger AI datasets
MedPMC's 11M curated medical image-text pairs trained a CLIP model beating larger datasets by 7.1 points — 95% of its images are actually medical.
Technology & AI
LLM hidden states predict answer confidence before generation
LLM hidden states hold richer confidence than models verbalise; a new method extracts it before generation finishes, affecting any AI system needing to know whe
Technology & AI
MCP server security: taint flaws mitigated via tool descriptions
Researchers propose SPELLSMITH, which rewrites MCP server tool descriptions to steer LLM agents away from taint-style vulnerabilities without code changes.
Technology & AI
ChatGPT clicks out in 5.2% of sessions, study finds
A new arXiv study finds ChatGPT sends users to external sites in just 5.2% of sessions, cutting Google search use 9.4% and reshaping the web's referral economy.
Technology & AI
SciReasoner AI model wins 67 of 86 science benchmarks
SciReasoner, a multimodal model for structural reasoning across proteins, molecules and crystals, matches frontier LLMs in 98% of expert reviews.
Technology & AI
Optimal control grows neural network layers where error peaks
A new arXiv preprint reframes neural network training as optimal control, inserting layers where error is highest — but evidence is limited to scientific datase