AI基础架构的批评
Gary Marcus指出当前LLM无法可靠地与数据库、知识图谱等基本工具协作。用户flowersslop补充LLM是被动反应式，通过添加循环和心跳（如openclaw）等补丁来解决根本问题。这种创可贴式方案掩盖了基础架构的缺陷。Marcus认为需要更坚实的AI基础。
Current AI is bandaids all the way down; LLMs can’t play nicely with basic tools like databases and ...
Current AI is bandaids all the way down; LLMs can’t play nicely with basic tools like databases and knowledge graphs, and you never know what you are going to get. It’s long time to face facts: LLMs have their uses, but we need a sounder foundation for AI. Flowers ☾ @flowersslop one problem I still have with LLM based AI is that its still reactive rather than proactive, so they add loops and stuff like the heartbeat in openclaw, but that pretty much feels like a bandaid solution for a way deeper more fundamental problem of how these things are built 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 4 👀 230 ⚡
- Clement Delangue06-01 21:28原文