数字速度下的科学：AI 对发现的影响

Science at Digital Speed: AI's Impact on Discovery

杰米斯·哈萨比斯 Demis Hassabis · 诺贝尔奖官方 · 2026-06-07 · 约 67 分钟 · 原视频 ↗

本期速览 · Overview

Demis Hassabis 探讨 AI 如何加速科学发现，从 AlphaFold 快速折叠蛋白质到智能体时代的曙光。

Demis Hassabis discusses how AI accelerates scientific discovery, from AlphaFold's rapid protein folding to the dawn of the agentic era.

要点 · TL;DR

AI 通过加速解决方案和即时传播来推动科学发现。
AI accelerates science by speeding solutions and enabling instant dissemination.
AI 将科学家从重复性任务中解放出来，促进创造性思考。
AI frees scientists from repetitive tasks, boosting creative thinking.
国际合作对 AI 安全至关重要，但地缘政治阻碍了这一点。
International collaboration is crucial for AI safety, but geopolitics hinders it.

核心观点 · Key points

AI 通过加快解决方案的速度和即时传播来加速科学发现。
AI accelerates scientific discovery by making solutions faster and disseminating them instantly.
AI 将科学家从重复性任务中解放出来，让他们有更多时间进行创造性思考。
AI frees scientists from repetitive tasks, allowing more time for creative thinking.
对齐问题尚未解决；确保 AGI 按预期行事至关重要。
The alignment problem is unsolved; ensuring AGI behaves as intended is critical.
AI 可以帮助科学家提出更好的问题并生成假设，而不仅仅是分析数据。
AI can help scientists ask better questions and generate hypotheses, not just analyze data.
AI 安全标准需要国际合作，但地缘政治使这变得复杂。
International collaboration is needed for AI safety standards, but geopolitics complicates it.

反共识 · Contrarian takes

生物学过于复杂，无法用简单定律描述；AI 作为描述语言比数学更合适。
Biology is too complex for simple laws; AI as a description language is better than math.
细胞是松散而非严格调控的；这种松散性可能防止系统陷入僵局。
Cells are sloppy, not tightly regulated; this sloppiness may prevent getting stuck.
意识和智能是可分离的；AGI 并不意味着意识。
Consciousness and intelligence are dissociable; AGI does not imply consciousness.
科学方法因领域而异；不存在单一的金标准。
The scientific method varies by field; there is no single gold standard.
AI 应该被少用而非多用，以保护人类的注意力和深度思考。
AI should be used less, not more, to protect human attention and deep thinking.
当前的 AI 热潮导致为收集数据而收集，缺乏理解；关注问题本身更重要。
Current AI hype leads to data collection without understanding; focus on questions matters.

本期章节 · Chapters（共 26）

数字速度下的科学 Science at digital speed
奇点与 AGI 时代 The singularity and AGI era
科学发现与接受的速度 Pace of scientific discovery and acceptance
AlphaGo：AI 的预警 AlphaGo as a warning shot for AI
学术奖励机制与放缓 Academia's reward system and slowing down
Demis 与 Paul 的渊源 Long history between Demis and Paul
湿科学家看 AI Wet scientist perspective on AI
AI 在实验室流程中 AI in lab workflows
AI 释放创造性思维时间 AI frees up time for creative thinking
AI 对科学创造与数据洪流的影响 Impact of AI on scientific creativity and data deluge
AI 解决现实问题 Applying AI to real-world problems
虚拟细胞模拟挑战 The virtual cell simulation challenge
一阶逻辑与学习系统的局限 Limitations of First-Order Logic and Learning Systems
模拟与粒度 Simulations and Granularity
计算资源与全球参与 Compute Access and Global Participation
给学术界的建议：聚焦 AI 理解与跨学科 Advice for Academia: Focus on AI Understanding and Interdisciplinary Work
担忧：滥用、AGI 风险与社会挑战 Worries: Misuse, AGI Risk, and Societal Challenges
国际合作与安全 International collaboration and safety
意识与可计算性 Consciousness and computability
哲学在科学中的作用 Role of philosophy in science
科学方法因领域而异 Scientific method varies across fields
AI 改善年轻研究者科研文化 AI improving research culture for young scientists
AI 是否改变了科学？ Has science changed with AI?
人类对游戏与机器的兴趣 Human interest in games vs machines
AI 作为科学工具 AI as a tool for science
乐观展望：发现的新黄金时代 Optimistic outlook: new golden age of discovery

阅读全文双语转录 →