Is there any Mandarin equivalent of the AGI Safety Fundamentals course? Someone could translate the curriculum into Mandarin. Translation doesn't matter as much if many Chinese people speak English, but that doesn't seem to be the case at all.

That's just one thought that motivated me to write this question. It would be extremely valuable to introduce Chinese students and professionals to AGI safety. Not only because China has a strong AI industry, but also because China has >1.4 billion people. Yet as far as I know, most AI alignment projects and organizations target English speakers. I've spent very little time researching AI alignment in China, and I could certainly be wrong.

If people want to do more research, I'd recommend the 2022 AI Index Report. Here is a (possibly misleading; again, I haven't looked into this carefully) graph from page 26:

From the 2022 AI Index Report.




New Answer
New Comment

4 Answers sorted by

Vael Gates's post "Resources I send to AI researchers about AI safety" offers this:

AI Safety in China

Both Human Compatible and the Alignment Newsletter have translations into Mandarin. There are also translations that are potentially less alignment-specific, like Life 3.0, The Precipice, Technical Countermeasures for Security Risks of Artificial General Intelligence, etc.

That's great. Seems that these days all the Alignment Newsletter translations go directly onto the English website.

Xiaohu Zhu
I have already made Alignment Newsletter sync now. 

Translation is a great idea.

It was one of the winners of the Future Fund’s Project Ideas Competition, and it's now listed on the project ideas page.

A problem unique to Chinese content is to ensure that it doesn't get blocked by their internet censorship policy.

Excellent, I'm happy to see that! However, I'm concerned that the proposal focuses entirely on translating general EA concepts.

Publications we might start with include, the 80,000 Hours ‘key idea’ series, and Toby Ord’s The Precipice.

I think it is much higher priority (from the perspective of reducing AI x-risk) to translate AI alignment concepts, particularly the AGI Safety Fundamentals course material. It takes a lot of inferences to go from "I'm interested in doing good" to "I like EA ideas" to "I think AI alignment is important" t... (read more)

Thanks for sharing these. The Chinese Association for AGI appears to focus on advancing AI capabilities rather than AI safety. I used Google Translate to translate the lead paragraph of the website's current opening page:

Notice of the 7th China General Artificial Intelligence Annual Conference

The China General Artificial Intelligence Annual Conference has been successfully held for six consecutive sessions. It is an annual event for Chinese general artificial intelligence enthusiasts, involving computer science, philosophy, logic, education, psychology, so

... (read more)
Sorted by Click to highlight new comments since: Today at 9:31 PM

Hi jskatt, great question! I’m a research analyst at Concordia and this is what I said in my Feb 2022 SERI talk re: AI alignment/safety-sympathetic resources/institutions in China:

“Over the past few years, Chinese researchers and policy stakeholders have demonstrated increasing interest in AI safety. 

For instance, last year two AI scientists from China’s AI Strategic Advisory Committee, which advises national policy on AI, wrote an article talking about the risks from AGI and potential countermeasures. The two scientists, Huang Tiejun and Gao Wen, along with their colleagues, present a summary of possible approaches to alignment based on Nick Bostrom’s book, Superintelligence, and cite other classic works in the Western AI alignment community like Concrete Problems in AI Safety and Life 3.0. The article acknowledged the relative lack of attention to AGI safety in China, and recommended “examining international discussions…of AGI policies, integrating cutting-edge legal and ethical findings, and exploring the elements of China’s AGI policymaking in a deeper and more timely manner.”

In the same year, Huang Tiejun and the chairperson of one of China’s top AI labs, the Beijing Academy of AI, endorsed the Chinese translation of Human Compatible, a book on AI alignment written by Professor Stuart Russell, who’ll be speaking at this conference tomorrow. Zhang Hongjiang also participated in a dialogue with Stuart at one of China’s most prestigious AI conferences, talking about the book and AGI safety.  

But despite these cases of high-profile support for AGI safety, many Chinese AI safety researchers focus on areas like robustness and interpretability instead of more alignment-relevant topics like goal specification.”

Separately, the org I work at – Concordia – aims to promote the safe and responsible development of AI, with a particular focus on China (more). For example, we recently wrapped up the first ever lecture series on AI alignment in China, which included speakers Rohin Shah, Max Tegmark, David Krueger, Paul Christiano, Brian Christian, and Jacob Steinhardt. To market this lecture series, we also translated a bunch of AI alignment works from English into Chinese. 

I’d be happy to have a chat about this, just messaged you.