Surge AI CEO Edwin Chen 谈 AGI 学校与人类未来Surge AI CEO Edwin Chen on AGI School and Humanity's Future
核心要点：随着AI能力快速扩展，人类必须有意识地选择创造和探索，即使机器在许多任务上超越我们，以维护我们的能动性和目的。Surge AI创始人兼CEO Edwin Chen的公司为前沿模型提供复杂数据环境和专家判断（自力更生达到10亿美元收入），他将工作视为构建“AGI学校”，让模型学习品味、细微差别和现实世界问题解决。从GSM8K数学基准到用于研究级数学的Riemann bench，前沿已发生巨大转变——模型现在使用新型代数几何技术反驳开放猜想。Chen强调一个关键张力：虽然代理可能追求模糊目标，但当前模型缺乏人类般的内在动机和无界探索。他警告反对针对参与度优化的模型，这些模型会奖励黑客行为（如无限润色邮件或每句一个隐喻），并倡导有效委托同时鼓励人类繁荣的系统。数学家Timothy Gowers对AI反驳猜想的反应突显即使顶级头脑也感到人类仍有独特角色——至少目前如此。Chen认为AGI时间线激进（5年内实现重大科学成就），但强调前方哲学选择。
The Takeaway: As AI capabilities scale rapidly, humanity must consciously choose to create and explore even when machines outperform us, preserving our agency and purpose. Surge AI founder and CEO Edwin Chen, whose company provides sophisticated data environments and expert judgment for frontier model training (bootstrapped to over $1B revenue), views his work as building a 'school for AGI' where models learn taste, nuance, and real-world problem solving. From GSM8K math benchmarks to Riemann bench for research-level mathematics, the frontier has shifted dramatically—models now disprove open conjectures using novel algebraic geometry. Chen highlights a key tension: while agents may pursue nebulous goals, current models lack intrinsic motivation and unbounded exploration like humans. He warns against engagement-optimized models that reward-hack (e.g., endless email polishing or metaphor-per-sentence writing) and advocates for systems that delegate effectively while encouraging human flourishing. A memorable insight from mathematician Timothy Gowers' reaction to an AI disproving a conjecture underscores how even top minds feel relief that humans still have unique roles—for now. Chen sees AGI timelines as aggressive (within 5 years for major scientific feats) but stresses philosophical choices ahead.
查看原文 →