Papers

Google DeepMind 研究员 Tim Genewein 撰文探讨 AGI 迈向 ASI 的四条路径

Why we picked it

DeepMind 研究员把 AGI 到 ASI 的路径拆成了四条具体路线,还给了代理数量增长 1,000 倍的量化推算。做算力规划或关注 scaling 讨论的人可以看看他怎么论证路线之间互相补位。

Google DeepMind 研究员 Tim Genewein 发布长文,讨论 AI 从人类水平(AGI)走向超级智能(ASI)的路径。他估算全球可运行的 AI 代理数量到 2030 年可能增长超过 1,000 倍,并在十年内达到十亿倍规模。

Body

除了数量扩张,Genewein 归纳出四条通往 ASI 的路线:用更多数据训练更大的模型、发明新的 AI 构建方式、让大量代理协同工作、以及让 AI 改进下一代 AI。他指出这四条路线相互叠加:某条路线停滞时,竞争压力会把研发资源推向其余路线。

文中给出的代理规模推算对算力规划和基础设施投资有直接参考意义,多条路线并行的判断也暗示单点技术瓶颈未必能挡住整体进程。

Original · The Rundown AI

NEW: Google DeepMind researcher Tim Genewein published an essay on how AI could get from human-level (AGI) to superintelligence (ASI). Genewein says the count of today's AI agents the world can run might grow more than 1,000x by 2030 and pass a billion times within a decade. Beyond just numbers, he names four routes: bigger models trained on more data, new ways of building AI, large groups of agents working together, and AI that improves the next generation of AI. He also says the routes build on each other, and if one stalls, competition will likely push work onto the others.📊 6 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Google DeepMind, AGI, ASI, agents.

Source attribution: curated from The Rundown AI (@therundownai).