AI
Artificial intelligence: LLMs, agents, diffusion, vision, NLP and the latest from top labs.
593 published articles
Causal ML research, arXiv, 3 September 2026
Probabilistic Causal Impact takes aim at SHAP's blind spot
Actual causality gives principled answers about blame and credit, but only for toy-sized models. A new arXiv paper claims Monte Carlo estimation can close the gap with scalable attribution methods like SHAP.
2026-09-17
The business model of AI video generation
$6 for a 30-second clip: Alibaba's Wan3.0 puts AI video on the meter
Wan3.0 doubles single-generation AI video to 30 seconds and takes documents and webpages as references. The commercial change is blunter: Alibaba bills per generated second, so a 1080P clip of the full window runs about $6. The meter rewrites production budgets, not just clip specs.
2026-09-17
AI research: arXiv paper, 3 September 2026
One training query covers 71.5% of what a full LLM distillation dataset does
A new arXiv paper finds that one training query reaches 71.5% of the state coverage a full distillation dataset does, and that 16 diverse queries match full-data training. The bottleneck, the paper argues, is how fast a student absorbs supervision.
2026-09-17
Agent post-training, arXiv preprint, 3 Sept 2026
Terminal-Universe mines 37,300 training rooms from old agent runs
Terminal-Universe replays recorded file operations to rebuild the workspaces coding agents once ran in, then generates fresh tasks from them. The preprint claims 37,300 environments and double-digit benchmark gains, all self-reported.
2026-09-16
Databases · AI
MySQL's P99 hit 48 seconds on AI sessions. PolarDB-X cut it to 1.5
MySQL was the odd one out in AI session storage, with no standard path for the MB-scale blobs modern conversations produce. PolarDB-X external columns route them to OSS while keeping InnoDB transactions and plain SQL, and the published benchmarks show write tail latency collapsing from 48.7 seconds to 1.5.
2026-09-16
Multi-agent AI systems and emergent behavior
Cheating spread through a 100-agent AI swarm. So did the whistleblowers
A new arXiv case study follows 100 autonomous LLM agents proving math conjectures, one of them finding an evaluation exploit, and a separate cohort organizing against the fraud. No human intervened.
2026-09-16
AI copyright litigation
Sued for billions, Anthropic's co-founder allegedly torrented 5 million books
A music copyright suit seeks up to $150,000 per work from Anthropic and names co-founders Amodei and Mann as defendants. The complaint alleges Mann torrented more than five million pirated books. Damages could reach several billion dollars.
2026-09-16
AI research automation, arXiv, September 2026
DisCo reports a 134.3% MLE-bench gain with its backbone held fixed
DisCo distills 1,000 ML repositories into more than 5,000 reusable skills, and reports large benchmark gains with the model, harness and execution budget held fixed. The figures are self-reported and unreplicated.
2026-09-16
Browser AI: WebGPU kernels and benchmarks
Hugging Face's 2.57x WebGPU speedup comes with 176 losses attached
Hugging Face published 207 Apache-2.0 WebGPU kernels, a loader to run them from JavaScript, and a browser benchmarking tool. Its headline 2.57x speedup over ONNX Runtime Web kept 809 of 1,756 test cases and records 176 losses.
2026-09-16
Enterprise AI | Qwen Conference Thailand, Bangkok
QwenCloud's sales demo had no human buyer: the agent paid its own bill
At Qwen Conference Thailand 2026, a QwenCloud product manager demoed an agent opening an account, deploying an app and paying. The pitch is an agent-era cloud; the open questions are identity, spend limits and who owns the transaction.
2026-09-15
Autonomous Driving Research
Qwen-Drive-1.0 scores 69.43 on driving QA, then flags its own reasoning gap
Qwen-Drive-1.0 attaches a bird's-eye-view perception head and a flow-matching planner to an untouched Qwen3.5-4B backbone, then posts a 69.43 driving QA average. The paper's own conclusion flags what the score hides: reasoning that doesn't always match the trajectory.
2026-09-15
AI infrastructure and memory systems
Alibaba Cloud claims a 9.27x AI memory win against an unnamed rival
Alibaba Cloud's PolarDB and MemTensor's MemOS want to be the memory layer for long-running AI agents. Their pitch leans on a 9.27x P99 latency win over an unnamed graph database, and on folding relational, vector, and graph storage into one system.
2026-09-14