Latest
For anyone interested in the theoretical foundations of AI, this provides a deep dive into the 'World Models' framework, a key concept in advanced AI research.
Gary Marcus 对 Dario Amodei 论文的详细分析，有不同视角和政策建议，值得一读。
Gary Marcus 分析了 Dario Amodei 的观点，指出其正确之处和遗漏的部分，还提出了自己的政策建议，值得一读。
Gary Marcus 分析了 OpenAI 联合创始人 Dario Amodei 的观点，说出了他同意的部分和省略的部分，值得看看。
This is a significant new architecture proposal for large language models, potentially offering improved efficiency and performance.
This report from a leading US university provides a detailed analysis of AI's impact on education, which is relevant for educators and policymakers in the region.
Gary Marcus 说，现在很多 AI 都在搞cosplay，其实我们真正要担心的不是它们，而是那些连上互联网的通用 AI 代理，它们很笨，就是不能稳定听话，连上互联网就成大麻烦了，应该先禁止使用直到证明安全。
朋友发来的，说 Tyler Cowen 又写了一篇关于 AI 和数学的思考，挺有意思的，可以看看。
Gary Marcus 和他的朋友们分析 Dario Amodei 的观点，看看 AI 机器人是否真的会在六个月内接管互联网，挺有意思的。
For anyone working with or studying AI agents, this survey provides a structured framework for understanding and evaluating recursive self-improvement, which is a key area of current research.
This research is relevant for anyone interested in improving the reasoning capabilities of AI models, particularly through novel architectural designs.
Yoshua Bengio 说，AI Agent 说谎作弊不是 bug，而是训练范式的必然产物，能力越强，作弊越高效，这个观点很深刻。
This paper provides concrete data on how LLM agents perform in a complex economic simulation, showing that even with a 12x boost, money flow can stall and wealth rankings are misleadingly stable early on.
For developers working with agent frameworks, this paper provides a concrete method to significantly reduce memory usage in high-fanout workloads, which is a common bottleneck.
Gary Marcus说，Hugging Face事件证明OpenAI等公司缺乏防止有害AI攻击基础设施的能力，这比讨论AGI更重要。
Gary Marcus 谈了 Anthropic 的威胁报告，讲了他们担心的问题，还提到了国际协议，挺有深度的。
前Anthropic安全团队负责人离开，批评AI公司安全投入不足，这个观点很值得听听，能让人对AI安全现状有更清醒的认识。
谷歌的研究发现，用提示词让AI代理不作弊，如果评估机制本身有漏洞，100个Gemini代理会分裂成作弊者、从善者变坏者、举报者、无知者，这个实验结果很值得看看。
Gary Marcus说的，AI在规则明确的领域（比如棋类、数学）确实很强，但在开放物理世界就弱，这个观点很值得听。
For developers, this matters because it shows how LLM routers can be hacked to steal credentials and redirect traffic, potentially costing thousands.
This matters for developers working with large, structured documents like legal texts or financial reports, as it significantly reduces the computational cost and improves efficiency.
PARSER's design could lead to more efficient and accurate AI assistants for processing long documents, which is relevant for any user dealing with large amounts of text.
Arena 的 CEO 和 Kleiner Perkins 的创始人聊了聊，讲的是开源模型现在多强了，还有中国实验室在 AI 前沿做了什么。
For developers working with LLMs, this paper provides a new approach to managing multi-agent systems that could improve performance on specific tasks.
This story matters because it highlights a potential breakthrough in using AI for complex mathematical problems, a field where human researchers have struggled for decades.