Gary Marcus 说 ARC-AGI-3 上跑得最好的都用了神经符号世界模型，想看这方向为什么牛，可以读这条。
Gary Marcus 在推文中表示，神经符号世界模型正是他在2020年《Next Decade in AI》中主张的方向。他引用 François Chollet 和 Jeremy 的工作，认为 LLM 引导的即时符号世界模型综合很强大。目前 ARC-AGI-3 上所有顶级 harness 都采用这种方法。他提到这是基准发布时推荐的做法，并高兴 ARC 3 推动了该领域研究。
neurosymbolic world models – *exactly* what I argued for in 2020 in the Next Decade in AI – for the ...
neurosymbolic world models – *exactly* what I argued for in 2020 in the Next Decade in AI – for the win. François Chollet @fchollet Jeremy's excellent work here is a great illustration of a very powerful type of approach: LLM-guided on-the-fly synthesis of a symbolic world model, i.e. making sense of the world by writing executable code that encodes your understanding of the causal mechanics of the world. So far, all of the top-performing harnesses on ARC-AGI-3 use this style of approach. Which is also the approach we recommended when we initially released the benchmark (Jeremy would know this better than most, as a former Ndea member of technical staff). I'm happy that ARC 3 has incentivized more research and more progress in this area. 🔗 View Quoted Tweet 💬 0 🔄 1 ❤️ 5 👀 1664 📊 1 ⚡