AI · SEMICONDUCTOR · FACT CHECK

SKD WIRE

EN WIRE · 2026-09-15 21:35

Chinese Researchers Unveil Five-Level Taxonomy for Self-Evolving AI, Placing Current Systems at Level 2

True

Chinese Researchers Unveil Five-Level Taxonomy for Self-Evolving AI, Placing Current Systems at Level 2
Image source: 조사 출처

A Chinese research team has published a classification scheme that sorts "self-evolving AI" into five developmental levels and judges the AI in use today to stand at Level 2. A verification by this newspaper, built on a review of nine primary sources, found the reporting to hold up, with every source consulted opening normally and without captcha, login, or access barriers.

원문 주장 (KR)
중국 연구진이 '스스로 진화하는 AI'를 5단계로 구분하는 분류법을 공개했으며, 현재 AI는 2단계 수준이라고 평가했다.

A five-stage ladder for AI that improves itself

The framework at the center of the claim divides self-evolving AI — systems capable of advancing on their own — into five distinct stages. Its central judgment is that today's AI occupies only the second of these levels, meaning current systems remain short of the higher stages of self-directed evolution that the taxonomy lays out.

How the check was done

Reporters went to the underlying materials directly rather than relying on secondary summaries. All nine primary sources reviewed during the fact-check were freely accessible, with no captchas, logins, or blocks standing in the way, allowing first-hand confirmation of both the five-level taxonomy and the researchers' own assessment of where current AI sits within it. Nothing in the sources reviewed contradicted the claim as it has circulated.

The claim is True.

Sources — primary documents (9)
  1. https://arxiv.org/abs/2609.11873
  2. https://arxiv.org/html/2609.11873v1
  3. https://theseus-labs-rsi.github.io/
  4. https://www.scmp.com/tech/tech-trends/article/3367486/chinese-researchers-chart-five-stage-path-toward-last-ai-built-humans
  5. https://theaiinsider.tech/2026/09/14/chinese-researchers-map-five-steps-toward-ai-that-can-improve-itself/
  6. https://www.aitimes.com/news/articleView.html?idxno=215284
  7. https://arxiv.org/html/2609.11873v1/rsi-overview-v11.png
  8. https://arxiv.org/html/2609.11873v1/L2-v2.png
  9. https://arxiv.org/html/2609.11873v1/Fig3loop_ver9.10.png

KR: /news/ · 판정: 사실