← 回總覽

定义 ASI 与 ARC-AGI-3 的结果

📅 2026-03-27 07:39 François Chollet 人工智能 2 分鐘 1585 字 評分: 82
ASI AGI ARC-AGI AI 能力
📌 一句话摘要 肖莱认为,解决 ARC-AGI-3 任务并不构成超级智能,因为 ASI 应该优于每一个人,类似于目前 AI 在国际象棋和围棋中的表现。 📝 详细摘要 在后续推文中,肖莱批评了将 ARC-AGI-3 的成功与人工超级智能 (ASI) 混为一谈的做法。他指出,如果普通人类都能解决这些任务,那么它并不代表这是一项不可逾越的挑战或超级智能。他将 ASI 定义为能够持续超越每一个人类的系统,并以国际象棋和围棋为例说明了现有系统。 📊 文章信息 AI 评分:82 来源:François Chollet(@fchollet) 作者:François Chollet 分类:人工智能 语言

Title: Defining ASI vs. ARC-AGI-3 Results | BestBlogs.dev

URL Source: https://www.bestblogs.dev/status/2037313406223704078

Published Time: 2026-03-26 23:39:02

Markdown Content: ![Image 1: François Chollet](https://www.bestblogs.dev/en/tweets?sourceId=SOURCE_fa42b4ed)

"2+ people can do it out of an unfiltered pool of 10 people that might well be a below-average sample" is not the sign of a insurmountable challenge. It's not certainly where I would set the bar for "super intelligence". ASI is when AI is better than every single human -- for instance we have ASI for chess and Go today.

4 Replies

1 Retweets

29 Likes

3,782 Views ![Image 2: François Chollet](https://www.bestblogs.dev/en/tweets?sourceid=fa42b4ed)

One Sentence Summary

Chollet argues that solving ARC-AGI-3 tasks does not constitute super intelligence, as ASI should outperform every human, similar to current AI performance in chess and Go.

Summary

In this follow-up, Chollet critiques the conflation of ARC-AGI-3 success with Artificial Super Intelligence (ASI). He posits that if average humans can solve these tasks, it does not represent an insurmountable challenge or super intelligence. He defines ASI as systems that consistently outperform every single human, citing chess and Go as existing examples.

AI Score

82

Influence Score 6

Published At Yesterday

Language

English

Tags

ASI

AGI

ARC-AGI

AI Capabilities

查看原文 → 發佈: 2026-03-27 07:39:02 收錄: 2026-03-27 10:00:40

🤖 問 AI

針對這篇文章提問,AI 會根據文章內容回答。按 Ctrl+Enter 送出。