Gary Marcus 在线怼人，翻出自己 2001 年的论文说对方没读过定义。想围观神经符号 AI 的考古级争论就看这条。
Gary Marcus 在 X 上吐槽，自己 2001 年就写过神经符号 AI 的原始定义，却被没读过的人教育。争议点是编码工具调用（coding harness）是否算神经符号 AI。Marcus 认为在 LLM 中嵌入并执行 WASM 更接近他 2003 年以来的相关实验。他贴出的链接指向 ChristosTzamos 的讨论。
OMG i wrote some of the original work on what is to be neurosymbolic in 2001 and dude who probably h...
OMG i wrote some of the original work on what is to be neurosymbolic in 2001 and dude who probably hasn’t read that work is trying to school me on the definition 🤦♂️ Grigori Karapetyan @GregKara6 coding harness and tools calls are not the original definition of neurosymbolic ai. you can change the definition or interpret it as such, but again was not the original interpretation. this, and many like it that came before it, attempt to do actual neurosymbolic ai by embedding and executing WASM directly in the LLM this is closer to your own definition of neuro-symbolic ai than any gymnastics to interpret harnesses as one. see here, and if you search for simillar experiments you can find them going back to 2003. 👇👇👇 x.com/ChristosTzamos… F5CH 🔗 View Quoted Tweet 💬 3 🔄 2 ❤️ 12 👀 2464 📊 4 ⚡