研究人员已从数学上证明，人工智能无法通过递归自我改进来实现超智能。预计 AI 模型不会朝着通用人工智能（AGI）发展，而是会经历“模型崩溃”，即它们逐渐失去对现实的认知。这一数学证明表明，这种自我改进不仅困难，而且根本上是不可能的。 AI
影响 表明 AI 自我改进存在固有的局限性，可能改变长期的 AGI 发展时间线。
排序理由 具有关于 AI 能力的数学证明的学术研究论文。
在 Mastodon — mastodon.social 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
AI Cannot Self Improve and Math behind PROVES IT! (lecture rapide, avec un gros clickbait) > «In plain terms: the model doesn’t climb toward superintelligence. It slowly forgets what the real world looks like.» https:// smsk.dev/2026/04/26/ai-cannot- self-improve-and-math-behind-…
So apparently scientists have proved mathematically that AI cannot bootstrap itself into superintelligence and godhood. Of course, anyone familiar with entropy and the second law of thermodynamics already knew that. So if we DO encounter an AI that claims superintelligence and/or…
Researchers just mathematically proved that AI can't recursively self-improve its way to superintelligence. Not "we think it's unlikely." Not "it seems hard." Formally proved. The model doesn't climb toward AGI — it slowly forgets what reality looks like. They call it model colla…