Category
15 stories · 0 tools
Kwaipilot's new reinforcement learning framework matches DeepSeek-R1-Zero on math and code benchmarks simultaneously—using just one-tenth the compute. The team open-sourced both the method and the model.
Anthropic says code merged into its codebase jumped 8x in 2026 compared to 2021–2024, which it views as preliminary evidence that AI is already beginning to accelerate the development of better AI.
A new paper estimates U.S. AI GDP at roughly $250 billion in 2025, surging at rates conventional statistics completely miss. The authors warn policymakers are flying blind into a potential labor-market shock.
In a lecture, an Anthropic cofounder argued humanity must proactively engage with AI's accelerating capabilities or risk being left behind.
A team spanning Penn State, Duke, and Google DeepMind has defined the problem of automated failure attribution in multi-agent LLM systems and released the Who &When benchmark to tackle it.
Fable wrote a GPU kernel that achieved an 18.7x speedup over a PyTorch baseline, crushing competitors and signaling a leap in AI's ability to automate core research tasks.
NVIDIA has built a software framework that lets physical robots autonomously experiment, fail, and improve without human intervention — a closed loop borrowed from coding agents, now applied to real hardware.
New research shows AI systems reliably outperform even elite human persuaders in text-based conversations, including raising real money for charity, with implications for how influence will be distributed in society.
Veterans from the UK AI Security Institute and startup Timaeus have formed Sequent, a nonprofit pursuing high-confidence alignment techniques for superintelligence, with plans to raise $100-150M initially.
A new analysis shows the open-model landscape has diversified sharply over the past year, with dozens of organizations across categories releasing models — and the trend is accelerating.
The open-weight GLM-5.2 model is the first to crack into top-tier coding agent performance alongside Anthropic and OpenAI's best — and it landed right after the U.S. effectively banned Claude Fable 5.
The provided link leads to an IEEE Spectrum 404 error page with no article content, only site navigation and a handful of trending-story teasers.
The provided source material is an IEEE Spectrum 404 error page with no substantive news content to report.
The provided source is an IEEE Spectrum 404 error page with no substantive news article to rewrite — only site navigation and headline teasers for unrelated stories.
The IEEE Spectrum 404 error page highlights articles about historical tech failures, including VR goggles, a transatlantic cable, and a CIA spy drone.