展望 2026 | sn 草稿

Looking Ahead to 2026

摘要 · Abstract

关于技术进步及其现实影响的笔记

Notes on advances in technology and real-world impact

提出 AI 应作为人类潜能的支架，而非替代品。
Proposes AI as scaffolding for human potential, not a substitute.
强调从模型向系统演进，协调多个智能体和工具。
Emphasizes evolution from models to systems with orchestrated agents and tools.
强调需审慎推广 AI 以应对社会挑战。
Stresses need for deliberate diffusion of AI to address societal challenges.
强调 AI 获得社会许可需有现实世界评估影响。
Highlights real-world eval impact for societal permission of AI.
呼吁建立包含认知放大器的新心智理论平衡。
Calls for new equilibrium in theory of mind with cognitive amplifiers.
指出 2026 年是 AI 从发现走向广泛扩散的关键年。
Notes 2026 as pivotal year for AI diffusion beyond initial discovery.

缺乏如何构建 AI 支架的具体技术细节。
Lacks specific technical details on how to build AI scaffolds.
未讨论推广可能带来的负面社会影响。
Does not address potential negative societal impacts of diffusion.
假设资源分配共识可实现，但未考虑现实冲突。
Assumes consensus on resource allocation is achievable.
对如何衡量现实世界评估影响的讨论有限。
Limited discussion on how to measure real-world eval impact.
侧重乐观视角，忽略滥用或不平等风险。
Focuses on optimistic view, ignoring risks of misuse or inequality.