简体中文
EnglishWhat Does AISafety.info Say About the Risks of Smarter-Than-Human AI?
Updated
AISafety.info argues that smarter-than-human AI may arrive soon and could lead to human extinction. Its core case rests on four claims: AI is advancing fast, AI may pursue goals of its own, AI can win a conflict against us, and experts are highly concerned. The site presents this as a summary of the mainstream concern rather than a single study, and it points readers to deeper arguments and ways to get involved.
The site's central claim
The homepage states the position directly: "Smarter-than-human AI may come soon. It could lead to human extinction." This is the framing for everything else on the site. It is not a claim that current AI systems are dangerous in themselves, but a claim about what happens once systems surpass human capabilities.
The four supporting arguments
AISafety.info lists four points under the heading "I need more arguments":
- AI is advancing fast. Rapid progress is the premise that makes the timeline short.
- AI may pursue goals. A system can act toward objectives rather than simply doing what a user intends.
- AI can win a conflict against us. If a capable system is in conflict with humanity, the site argues the system can prevail.
- Experts are highly concerned. The site treats expert concern as evidence that the risk deserves attention.
Each point is a link to further material, so the homepage functions as an entry point rather than a full argument.
How the site is organized
The content is split into levels so you can choose your depth:
| Section | What it contains |
|---|---|
| Basic sections | Introductory material on the core risk claims |
| Advanced sections | More detailed arguments and analysis |
| Browse all articles | The full article list |
| Browse by category | Articles grouped by topic |
If you only want the short version, the homepage claims are enough. If you want to evaluate the reasoning, the advanced sections and the article index are where the actual arguments live.
Asking questions directly
The site offers Stampy, a chatbot for AI safety questions. You can type a question into the prompt on the page. The site itself warns: "Stampy can be inaccurate. Always verify its sources." Treat its answers as a starting point for finding sources, not as a citable conclusion.
Who runs it and how to get involved
AISafety.info is a project founded by Rob Miles and maintained by a global team of specialists and volunteers. The site describes its purpose as ensuring the effects of future AI are beneficial rather than catastrophic. It is an Ashgro Inc project; Ashgro Inc is listed as a 501(c)(3) public charity incorporated in Delaware, with EIN 88-4232889. The footer carries a copyright range of 2022–2026.
Ways to participate, as listed on the site:
- Get involved — general participation
- Donate — financial support
- Code — contribute to the project
- Write — contribute written material
- Join us on Discord — community discussion
The site also lists partner projects: AISafety.com and Alignment Ecosystem Development.
What this means for you
If you want a quick statement of the AI-extinction argument, the homepage gives you the four claims in a few lines. If you want to check whether those claims hold up, follow the links into the advanced sections and read the underlying articles. If you have a specific question, Stampy can point you toward material, but verify anything it says against the cited sources. If you want to contribute rather than just read, the code, writing, and Discord routes are the concrete options the site offers.