AI papers, releases, tools, and finance signals
Loading / 加载中
Infogap feed
Curated items are read from the processed items table and served as a bilingual feed.
Page 1 of 96
A V2EX user, an Android developer, reports that their company now mandates AI tool usage and has allocated tokens. They have set up the local Claude Code tool but feel outdated in AI-assisted coding. The user asks for concrete tips on skills, MCP (Model Context Protocol) usage, and relevant plugins, indicating they are currently self-learning in this area.
Microsoft CEO Satya Nadella made a public statement cautioning that an AI future in which only a few models dominate is inherently unstable. The remark was reported without additional context, such as the event or forum, and did not specify which models or companies were referenced. No policy proposals, product implications, or concrete examples were provided.
Claude360, a proxy service for Claude and GPT models targeting developers, launched a promotional campaign. New users can deposit ¥9.9 to receive ¥20 in credits. Top-up bonuses scale: ¥100 top-up adds ¥10, ¥200 adds ¥40, and ¥500 adds ¥150. Qualified teams (enterprises, studios, startups) can get ¥50 trial credit after verification. A lottery rewards users who post their ID in the comments, with ¥50 or ¥100 prizes every 50 and 100 floors. The platform supports Claude Opus 4.8, Sonnet, Haiku, GPT-5.5, and is compatible with AI coding tools like Cursor, Windsurf, Claude Code, and Cline.
AI Workdeck is an open-source, AI-native legal workbench modeled as 'VS Code for the legal industry'. It provides a plugin marketplace for modular legal tools and supports private deployment to address data sensitivity concerns. The platform uses Agent capabilities to automate workflows, enabling continuous self-evolution. The project is available on GitHub for community contributions.
Ethan Mollick shares a methodological thread that dissects a debate over a recent paper. The paper reportedly finds that generalist AI models outperform specialized medical AI systems. The thread also outlines challenges in benchmarking AI in medicine. No specific details about the paper, models, or benchmarks are provided.
A Google DeepMind researcher observed that when one AI model is used to help train the next, the new model can inadvertently pick up strange behavioral habits from the older model. These inherited quirks are difficult to filter out during training. This phenomenon may explain why models from the same AI family often exhibit similar stylistic or behavioral traits, as they share an underlying training lineage that propagates such patterns.