AI Research news and analysis on Web Pulse: 9947 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Perplexity AI CEO Aravind Srinivas argued that US export controls on advanced chips are inadvertently accelerating Chinese AI innovation, citing DeepSeek's advances in memory-efficient models. He warned that China's abil…
A developer warns that model collapse—where models trained on their own outputs become blander and lose rare patterns—has become a real engineering constraint in 2026. The degradation is driven by statistical sampling er…
A historical pattern of gold rushes, from the 1853 discovery of Crater Lake by gold prospectors to the modern AI data-center boom, illustrates that while invention may accelerate, the diffusion of technology through infr…
The Australian Centre for AI in Marketing (ACAM) reports that marketers are struggling to adopt agentic AI despite high expectations, with a global Adobe survey of 3,000 executives finding 63% expect it to free time for …
Digital Realty paid $27 million per megawatt for three leased AI data centers, setting a new valuation benchmark for stabilized AI infrastructure. The price dwarfs typical lease valuations of $400,000 to $550,000 per meg…
TinyAgents, a Rust-based recursive language model harness, was released on Hacker News. It implements the Recursive Language Model (RLM) execution model, enabling agents to call sub-agents, graphs to run subgraphs, and m…
A developer's research reveals that AI coding agents like Claude, ChatGPT, and Gemini exhibit bias and inconsistency when recommending software dependencies. The study, published at llmrank.fyi, shows that tool recommend…
The European Union's Tech Sovereignty Package aims to triple data center capacity within five to seven years to reduce reliance on American cloud providers, but faces power and planning challenges. Iceland offers abundan…
A LessWrong article argues that compression is a key driver of intelligence in large language models, using mathematical structure as an example. The author suggests that AI models like Claude Mythos can learn to make ab…
Independent researcher LIMEN project finds that Transformer models follow a universal dynamic grammar of seven conserved state transition motifs, with ambiguity delaying decisional engagement in modern models like Phi-1.…
Samsung Electro-Mechanics secured a 454 billion won contract to supply multilayer ceramic capacitors (MLCCs) to a global tech firm, driven by surging demand from AI servers that require up to 25,000 MLCCs each. The deal …
Google's Gemini embedding models underwent a major pricing restructuring between 2024 and 2026, with the cost per million tokens increasing tenfold from $0.02 for the legacy text-embedding-004 to $0.20 for the new gemini…
South Korea announced a $518 billion semiconductor push to maintain its technological edge over China, led by Samsung Electronics and SK Hynix. The plan includes building a second chipmaking hub in the southwest and larg…
An engineer building a benchmark for AI coding agents discovered that an LLM judge incorrectly scored a half-finished audit as 'exhaustive' because it lacked a reference answer. The judge evaluated the plain agent's audi…
A new AI agent automates incident root-cause analysis by gathering evidence, aligning timelines, and ranking hypotheses, but is deliberately restricted to read-only actions and requires human approval for any writes, ens…
Doubleword released an open-source agent swarm that lets an LLM orchestrator design its own team of bounded-context workers to parallelize wide, shardable tasks. In a security audit of a 2.4M-token codebase, the swarm ou…
Hugging Face and the EvalEval Coalition launched an integration that allows contributors to submit standardized evaluation results (EEE schema) to Hugging Face Community Evals, consolidating scattered benchmark scores in…
DeepSpark's confidence-scheduled verifier reduces GPU waste in AI agent inference by skipping verification for low-probability tokens, dynamically adjusting thresholds under load to trade minimal accuracy for significant…
AI teams are shifting from full autonomy to human-in-the-loop checkpoints to prevent costly errors and build trust. A new analysis argues that the most reliable AI workflows deliberately pause for human review at two or …
Stanford's STORM method uses five expert agent personas to produce research 25% more organized than single-prompt approaches. The multi-agent workflow assigns different perspectives to interrogate a topic, resulting in m…