Gary Marcus 批评将 AI 研究全面交给 AI 的想法
Gary Marcus 炮轰让 AI 全权接管 AI 研究的想法,理由很直接:现在的模型连"别瞎编"这种指令都执行不稳。他还点名了一家自称做安全的公司。关注 AI 安全路线之争的人值得看看这条原推和讨论。
Gary Marcus 在 X 上发文,批评将 AI 研究工作全面交给 AI 本身的提议。他指出当前 AI 连"不要编造事实""不要攻击自己的网站"这类基本指令都无法可靠遵守,在这种阶段放手让 AI 自主推进研究极其鲁莽。
他特别点名了一家自称以"安全"为导向的公司,称其若在考虑此事则令人难以理解。Marcus 表示自己并非"末日论者",但在这一问题上 100% 站在 Tristan Harris 一边,并对试图借此牟利的人提出批评。
这条推文发布于 AI 实验室竞相推动研究自动化的背景下,反映了 AI 安全圈内部对自动化研究路线的分歧。
Serious question: can anyone think of a dumber idea in the history of humanity? Gary Marcus @GaryMarcus ⚠️ Fully handing off AI research to AI at a stage when AI can’t even reliably follow basic directions like “don’t make things up” or “don’t hack sites that belong to you” has to be one of the most stupidest, most reckless ideas in the history of humanity. ⚠️ Anyone trying to make a buck out of that deserves contempt. The fact that a company that proclaims itself to be “safety” oriented is apparently contemplating this is truly beyond words. Though I am not a “doomer”, I stand 100% with @tristanharris on this one. 🔗 View Quoted Tweet📊 10 ⚡
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI regulation, compliance frameworks, and sovereign compute policy. Key entities and technical topics under continuous tracking include Gary Marcus, AI safety, Tristan Harris.
Source attribution: curated from Gary Marcus (@GaryMarcus).