AI Research news and analysis on Web Pulse: 22681 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Mechanistic interpretability researchers have abandoned the search for a single "One True Technique" after many-to-many neuron-to-concept mappings discovered in 2023 failed to transfer from artificially-simple 512-neuron…
A developer trained a 3.8-billion-parameter dense language model to a CORE score of 0.384 for a total budget of $998, according to an account of the Little LM project. The run relied on Grouped Query Attention to cut KV-…
Researchers introduced PlannerForge, a system that uses LLM agents to automate scenario-based testing of motion planners in autonomous driving, according to the paper's description. Scenario-based testing for Autonomous …
Fetch.ai's Innovation Lab has deployed custom autonomous AI agents at San José State University and California State University, Long Beach, built on its uAgents framework, Agentverse platform, and ASI:One, with plans to…
Sumama Jameel, a 14-year-old developer in Karachi, Pakistan, has shipped nine products that he designed and built by directing AI to write, test, and debug the code. His primary project, Megabit, is a data generation eng…
A developer's analysis of the aiappdex.com model catalog highlights three non-text AI models seeing heavy production use: Amazon's Chronos-2 for zero-shot time-series forecasting (25.7M downloads), OpenAI's CLIP-vit-base…
Anthropic researcher quits over out-of-control AI fears, drawing calls for new AI safety rules from dozens of politicians in the U.S. and abroad, as the Wall Street Journal reported. The essay examines why AI lab employe…
A survey examines inference-efficiency mechanisms in video and audiovisual large language models (VideoLLMs), which pair video representations with pretrained large language models and condition generation on a textual p…
Mathematicians Tristan Buckmaster, Levent Alpöge, and Matei Coiculescu published machine-checked Lean proofs showing that several fluid equations, including 3D incompressible Euler, can blow up in finite time under smoot…
DeepSeek released its V4.1 Flash model on Thursday, claiming it outperforms its previous flagship V4 Pro on coding, cybersecurity and autonomous agent benchmarks while cutting inference costs. The model uses a new "Causa…
A study accepted at GAISS 2026 found that reviewing AI-generated code against an approved specification baseline did not help reviewers catch more bugs, but made the bugs they did catch contract-anchored, attributable, a…
Research presented at NeurIPS 2025 from the MAST taxonomy, developed by UC Berkeley researchers, found that 79% of multi-agent system failures stem from specification and coordination problems rather than model capabilit…
Anthropic researchers Anton Korinek, Charles I. Jones, Szymon Sacher, Tess Cotter, and Peter McCrory published a framework modeling AI's economic effects between 2026 and 2030, with three scenarios: modest, substantial, …
A developer measured how the claude-sonnet-5 model answers company-register questions with and without access to a new MCP tool called registry-mcp, which unifies Norway's Brønnøysundregistrene, the UK's Companies House,…
Jacob Coxon, a 27-year-old British mathematician and AI researcher, resigned from Anthropic on 9 September after roughly four months and warned that AI 'could kill us all by the end of the decade,' a claim that went vira…
DeepSeek AI released DeepSeek-V4.1-Flash, a multimodal Mixture-of-Experts model with 552B backbone parameters, 196B additional Engram parameters, and a 1M-token context window, cutting global KV cache footprint to 890 by…
A software developer argues that AI coding tools have already surpassed most human developers at shipping code, citing research showing frontier AI agents can complete tasks that would take humans days or weeks. The deve…
DeepSeek released V4.1-Flash, an efficiency-focused model with 552 billion total parameters (8 billion active for input, 16 billion for output) that matches or beats GPT-5.6 Sol and Claude Opus 5 on several agentic and c…
A technical post in the Machine Learning series demonstrates teaching the sw-MLPL array language to express convolutional neural network operations, using the triple-sum convolution definition from Zhao, Wang, Wang and L…
Aiden AI published a dev-board-stage architecture writeup describing how it splits a full-duplex voice agent into a real-time voice model and a separate, stronger background model that coordinate asynchronously through a…