AI Research news and analysis on Web Pulse: 10765 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Researchers at Arizona State University argue that calling intermediate tokens generated by language models 'reasoning traces' or 'thinking traces' anthropomorphizes the models and misleads users about their capabilities…
Samsung Group plans to invest $646 billion over the next decade in semiconductors and AI data centers, marking one of the largest corporate investment commitments in South Korea's history. The investment includes 300 tri…
Aaron Levie argues that de facto AI regulation has arrived, with powerful models likely requiring government review before release. He warns this could slow innovation, create release backlogs, and incentivize sovereign …
The Command and General Staff College (CGSC) is integrating artificial intelligence into its training for the FY25 school year to achieve training objectives at echelon, reflecting the Army's pursuit of technological sup…
Quantinuum announced Helios, a 98-qubit trapped-ion quantum computer with record-low error rates, as published in Nature. The system achieves single-qubit gate error rates of 2.5 in 100,000 and two-qubit gate error rates…
Unconventional AI released Un-0, an image generation model powered by coupled Kuramoto oscillators instead of conventional neural network architectures, publishing code and weights to demonstrate its physics-first comput…
Photoroom engineers found that classifier-free guidance scales above 7.5 caused oversaturation and clipped highlights in their SDXL product renders. They fixed the issue by adding CFG rescale at 0.7 and dynamic threshold…
Altus Schools in San Diego spent $500,000 on two AI-powered humanoid robots named Ameca for a pilot program exploring AI in education. Critics argue there is no evidence these tools are effective or safe, and they may ca…
Researchers from Microsoft Research, UC Berkeley, UC San Francisco, and Columbia University developed generative causal testing (GCT), a method that distills black-box brain-prediction models into testable verbal explana…
DeepSeek, an AI research company, is actively recruiting for various positions as indicated by its job listings on Mokahr. The company seeks to expand its team to advance its work in artificial intelligence.
Researchers have published a paper arguing that the design principles underlying MLIR, a compiler infrastructure for AI models, have formal theoretical correspondences that can clarify abstraction completeness and guide …
Ollama's expansion to support Chinese AI models like Kimi-K2.5, GLM-5, MiniMax, and DeepSeek offers local deployment benefits but comes with hidden costs. Developers face documentation gaps, quantization trade-offs, and …
Frontiers research integrity manager Simone Ragavooloo argues that the research community must stop conflating all AI-related issues with misconduct, as a new whitepaper reveals inconsistent definitions and polarized vie…
Google's WebMCP, now in public Origin Trial in Chrome 149, lets developers expose structured tools to in-browser AI agents, reducing token usage by 90% compared to screenshot-based automation. The open web standard, co-d…
Tuya Smart co-founder Alex Yang told the Summer Davos forum that China's rapid AI adoption is driven by a vast user base, openness to new technology, and societal aspiration for progress. Yang highlighted Tuya's AI appli…
Asian technology stocks slumped after Apple raised prices for its products, stoking concern that rising component prices will curb demand for devices and slow the memory chip rally. South Korean memory chip giants SK Hyn…
AI teams often hit a 'latent capability ceiling' where larger models no longer improve task accuracy, with scaling data showing diminishing returns beyond 100B parameters and 61% of tasks exhibiting non-monotonic scaling…
A new open-source repository, the Agent Engineering Roadmap, provides a structured, beginner-friendly guide for building production-ready AI agents, covering topics from single agents to multi-agent colonies and producti…
A 428-billion-parameter open-weight model achieved 59.0% on SWE-Bench Pro, outperforming GPT-5.5's 58.6%, after being compressed from 855GB to 128GB. The model's efficiency and performance mark a significant advance in A…
Researchers at arXiv find that verifying coding agent outputs is now harder than generating them, as foundation models improve. They argue that no fixed reward function remains effective as policy capability grows, and v…