Gary Marcus 的冷静判断值得 AI 从业者关注——他指出了 LLM 的局限性并提供了长期路线图，做 AI 战略或技术选型的人看完会有感触。
AI 学者 Gary Marcus 在 X 上发文，认为 LLM 虽然有用，但距离真正的人工智能还有很长的路要走。他推荐自己 2020 年的文章《The Next Decade in Arxiv》，称其仍然是未来的良好指南，并且公司们正越来越多地遵循其中的路线图（尽管没有公开承认）。Marcus 回应了关于 LLM 是否只是拼图的一小块的质疑，暗示当前路径可能并非终点。
LLMs are useful but we still have a long way to go. See my 2020 article The Next Decade in Arxiv, wh...
LLMs are useful but we still have a long way to go. See my 2020 article The Next Decade in Arxiv, which remains a good guide to the future and offers a road map that the companies are increasingly following (without really saying so out loud). FIONG @FI0NG @GaryMarcus @dread_numen I know your big skeptic of Gen Ai but do you think LLMs are one small piece of puzzle and we just need to figure out rest? Or is it totally path that is short end 🔗 View Quoted Tweet 💬 3 🔄 0 ❤️ 6 👀 1017 📊 2 ⚡