马克·扎克伯格、普莉希拉·陈和亚历克斯·里夫斯讨论 Biohub 如何利用开源工具和 AI 对蛋白质、细胞及整个生物系统进行建模，以加速科学进步。
Mark Zuckerberg, Priscilla Chan, and Alex Reeves discuss how Biohub uses open-source tools and AI to model proteins, cells, and whole biological systems, aiming to accelerate scientific progress.
要点 · TL;DR
AI 能将生物学从发现科学转变为工程科学。 AI can turn biology from discovery-based to engineering-based science.
构建从蛋白质到细胞的层级世界模型是关键。 Building hierarchical world models from proteins to cells is key.
开源工具比专有工具更能加速科学进步。 Open-source tools accelerate scientific progress more than proprietary ones.
核心观点 · Key points
AI 能将生物学从发现科学转变为工程科学。 AI can transform biology from discovery-based to engineering-based science.
构建从蛋白质到细胞再到整个系统的分层世界模型是关键。 Building hierarchical world models from proteins to cells to whole systems is key.
开源工具比专有工具更能加速科学进步。 Open-source tools accelerate scientific progress more than proprietary ones.
前沿 AI 和前沿生物学必须在一个组织内紧密整合。 Frontier AI and frontier biology must be tightly integrated in one organization.
模型的可解释性可以揭示未知的生物学知识。 Mechanistic interpretability of models can reveal unknown biology.
反共识 · Contrarian takes
在本世纪末治愈所有疾病的目标现在被认为过于保守。 Curing all diseases by end of century is now seen as too conservative.
AI 模型无需专门训练就能设计蛋白质。 AI models can design proteins without being explicitly trained for that task.
罕见病可能比常见病更快解决，因为患者驱动的试验更高效。 Rare diseases may be solved faster than common ones due to patient-driven trials.
一小群 AI 研究人员就能在生物学上取得重大突破。 A small team of AI researchers can achieve major breakthroughs in biology.
最大的瓶颈是生成新的生物学数据，而不是算力。 The biggest bottleneck is generating novel biological data, not compute.
本期章节 · Chapters（共 22）
引言与使命Introduction and Mission
Biohub的起源与理念Origin and Philosophy of Biohub
Biohub与虚拟生物学计划的演进Evolution of Biohub and Virtual Biology Initiative
从单细胞测序到AI驱动生物学From Single-Cell Sequencing to AI-Driven Biology
连接还原论与系统生物学Bridging reductionist and systems biology
非营利模式与风投公司对比Nonprofit approach vs venture-backed company
构建前沿生物学模型的挑战Challenges in building frontier biology models
首先受影响的疾病领域预测Predictions on disease areas impacted first
科学进展的时间表Timeline for scientific progress
临床转化的挑战Clinical translation challenges
ESM Fold发布与蛋白质世界模型ESM Fold launch and protein world model
实验室中的分子表征Characterizing molecules in the lab
开放生态系统与赋能个人Open Ecosystems and Empowering Individuals
吸引人才加入BiohubAttracting Talent to Biohub
从蛋白质模型到临床影响From protein models to clinical impact
发布以来的激动应用Exciting uses since release
决定下一步研究方向Deciding next research steps
AI带来的兴奋与疲惫Excitement and exhaustion in AI
生物学世界模型的五年愿景Five-year vision for biology world models