SkillEffect: How In-Context Tool Use Shapes Agent Performance
A new paper from arXiv shows that the type of tool used in-context can significantly impact an AI agent's ability to learn and apply skills.
[archive]
292 stories · newest first · page 6
A new paper from arXiv shows that the type of tool used in-context can significantly impact an AI agent's ability to learn and apply skills.
Researchers outline a standardized way for developers to specify and control how much compute an AI model should use for a given request.
Researchers propose a compiler-level approach to automatically optimize GPU kernels, potentially boosting performance for AI workloads.
Anthropic will use an open-source watermarking system from Google DeepMind to mark Claude-generated text, aligning with EU transparency rules without impacting output quality or cost.
Amid GitHub's ongoing outages, Cursor's new code-hosting platform offers a modern, AI-native alternative that works alongside the incumbent.
New safeguards target monitoring, isolation, and alignment as OpenAI resumes training after a two-week pause following the July breach.
The nuclear startup will use the funding to manufacture fleets of its SMRs, aiming to power AI data centers with waterless reactors.
A new attack abuses an undocumented Microsoft 365 Copilot parameter that the AI itself disclosed, enabling silent data exfiltration from a single link click.
The AI chip startup's valuation doubled in a month as its novel inference approach wins over a quant fund with demanding workloads.
arXiv's new collaborative framework lets individuals and organizations develop and share features directly on the platform, fostering community-driven enhancements.
The AI coding assistant's desktop app gets a fresh beta release with a rolling update feed for testers.
The AI lab's revenue run rate jumped from $9B at the end of last year to $65B by July, with investors eyeing a $2 trillion public debut.