T
traeai
登录
返回首页
Gary Marcus(@GaryMarcus)

If we had done everything I suggested in my 2020 arXiv article “The Next Decade in AI”, we might act...

8.5Score
If we had done everything I suggested in my 2020 arXiv article “The Next Decade in AI”, we might act...

TL;DR · AI 摘要

Gary Marcus认为,如果在2020年提出的AI十年规划中所有建议都得到实施,我们可能已经达到了AGI。然而,过度依赖纯扩展导致其他关键目标进展缓慢。

核心要点

  • Gary Marcus在2020年的arXiv文章中提出了AI十年规划。
  • 目前,AI在神经符号AI方面取得了进展,但在知识数据库、可靠推理系统和世界模型表示方面进展缓慢。
  • 过度依赖纯扩展延迟了AI的全面发展,未来应专注于其他三个关键目标以实现AGI。

结构提纲

按章节快速跳转。

  1. Gary Marcus回顾了2020年提出的AI十年规划,并指出如果所有建议都得到实施,可能已经达到了AGI

  2. AI在神经符号AI方面取得了进展,但其他关键目标进展缓慢。

  3. 过度依赖纯扩展导致其他关键目标进展缓慢。

  4. 未来应专注于其他三个关键目标以实现AGI。

思维导图

用一张图看清主题之间的关系。

查看大纲文本(无障碍 / 无 JS 友好)
  • AI十年规划

金句 / Highlights

值得收藏与分享的关键句。

  • 如果在2020年提出的AI十年规划中所有建议都得到实施,我们可能已经达到了AGI。

    第1段

    ⬇︎ 下载 PNG𝕏 分享到 X
  • 目前,AI在神经符号AI方面取得了进展,但在知识数据库、可靠推理系统和世界模型表示方面进展缓慢。

    第2段

    ⬇︎ 下载 PNG𝕏 分享到 X
  • 过度依赖纯扩展延迟了AI的全面发展,未来应专注于其他三个关键目标以实现AGI。

    第3段

    ⬇︎ 下载 PNG𝕏 分享到 X
#AI#AGI#Gary Marcus#神经符号AI#知识数据库
打开原文

In the last three years, after a detour driven by the false promise of pure scaling, we have gone a long way to the first of the four goals I laid" / X

Gary Marcus, MIT PhD and NYU Professor Emeritus on X: "If we had done everything I suggested in my 2020 arXiv article “The Next Decade in AI”, we might actually have reached AGI by now. In the last three years, after a detour driven by the false promise of pure scaling, we have gone a long way to the first of the four goals I laid" / X

Don’t miss what’s happening

Image 1

Gary Marcus, MIT PhD and NYU Professor Emeritus

@GaryMarcus

If we had done everything I suggested in my 2020 arXiv article “The Next Decade in AI”, we might actually have reached AGI by now. In the last three years, after a detour driven by the false promise of pure scaling, we have gone a long way to the first of the four goals I laid out: neurosymbolic AI. (Claude Code; DeepMind’s new math system that solved 9 Erdos problems, etc). Harnesses and tools, for example. But the field has made little progress on the other three: - databases of explicit, systematic, machine interpretable knowledge [LLMs are a partial but inadequate/unreliably substitute] - reliable reasoning systems that can work with partial and incomplete information - the ability to represent and induce explicit world models (which I called cognitive models in 2020) that AI systems can reason over. An excess commitment to pure scaling delayed all of these. If we focus on the other three AGI may be achievable, possibly in the next decade.

Last edited Opens edit history 4:04 PM · May 27, 2026

·

11.1K Views

27

19

73 77

35

Read 27 replies

AI 可能会生成不准确的信息,请核实重要内容