Marcus 的批评戳破了 AGI 已来的乐观叙事，关注 AI 能力边界的开发者或研究者值得一看——它提醒我们不要被炒作迷惑，重新审视当前模型的实际局限。
Gary Marcus 在 X 上回应 Google DeepMind 的 Oriol Vinyals 关于“AGI 已以某种形式到来”的观点，强调 AGI 远未实现。他引用自己与 Miles Brundage 的赌约中的十个例子，认为当前没有任何 AI 能可靠完成其中任何一个，更不用说一个通用系统完成全部。Marcus 指出，AGI 的定义标准在不断提升，但当前系统仍远未达到他、Dan Hendrycks、Yoshua Bengio 等人在 agidefinition.AI 上提出的严格标准。这场争论反映了 AI 社区对 AGI 进展的不同评估视角。
AGI is certainly not here by the definitions I have repeatedly laid out. or by criteria that @hend...
AGI is certainly not here by the definitions I have repeatedly laid out. or by criteria that @hendrycks @Yoshua_Bengio and I and others recently laid about at agidefinition.AI i don’t think any current AI can even reliably do any of the ten examples in my bet with @miles_brundage , let alone a single (general) system doing all of them. Haider. @haider1 Google DeepMind's Oriol Vinyals: "AGI is already here in some way, by the definitions we used a few years ago" But the expectation keeps moving, while today's systems have already passed what many expected a few years ago It is not here in the ideal form yet, but it is getting close Your browser does not support the video tag. 🔗 View on Twitter 🔗 View Quoted Tweet 💬 12 🔄 1 ❤️ 23 👀 2818 📊 11 ⚡