Hide table of contents

Sam Altman (CEO of OpenAI), Demis Hassabis (CEO of DeepMind), and Eric Schmidt (ex-CEO of Google) have all publicly taken a very hawkish stand in the US-China geopolitical relationship. Western elite in general seem to be moving towards being much more hawkish towards China. They see China as the enemy the West needs to overcome.

China is emerging as the preeminent world superpower, and will likely be a larger economic powerhouse than the US within our lifetimes. 

Rather than see this as a competition, the Western elite could and should try to push for collaboration through increased global trade, diplomacy, and mutual respect of territorial and other geopolitical desires. If we haven't had a World War since WW2 because of the proliferation of global trade, then wouldn't it be reasonable to think that a reduction in global trade, as the Chinese economy focuses inwards and the West disengages with it, puts the entire world into an extremely precarious position?


World War, today

A world war fought with modern technology will have catastrophic effects on the lives of regular people, and could be a major x-risk. Military technology progressed very quickly during WW1 and WW2, as government resources were brought to bear upon the problem of killing many people as cost effectively as possible. Modern military technology is extremely outdated, but would very quickly advance if there was a world war. For example, we will likely see lots of killer robots and killer drones: for a few $100, enough to kill 10s-100s of civilians. Cars could be hacked en-masse. Nuclear missiles would target large urban population centres.


Influential tech people

Tech people are starting to gain a lot of influence in national security/defence circles within the US. If you're in tech, and you have peers who are like this, you probably have seen the phenomenon I'm describing. All of the tech people see China as the enemy. China doesn't need to be the enemy! Until now, we've had a great economic relationship: and we should keep it up! The alternative is a large world war.

If you have any friends like that, please talk to them about this and try to convince them to tone it down. 

Consider OpenAI. OpenAI thinks it will be able to develop safe AGI and advance Western values. The subtext here is to subjugate the people of China to be ruled by Western values. Doesn't this framing seem to amplify conflict, rather than contain it? If you were the Chinese government, wouldn't this concern you and push you towards responding in kind?


Action plan

I think EAs should get on the same page that a collaborative US-China relationship is something we should strive for. It's a very important factor in our long term future. 

If we can get on the same page about this, it would be great to understand what kinds of work could be pursued by young professionals in this cause area.

Why isn't this view already quite common if it's so obvious? Because internationalism, and the equal value of all human beings is not a universally shared value, especially among the national security elite of the United States. They'd rather see the US do great and the rest of the world suffer, than to see the US fall behind and the rest of the world develop further.

And it seems like our tech friends are starting to go along with this view. I hope we can change their minds.





More posts like this

Sorted by Click to highlight new comments since: Today at 4:27 PM

Before I start tearing into this post, I just want to clarify that 1) the mistakes that this post makes are ones that are impossible for highly competent people to avoid, it is instrumentally convergent for EAs to worry about China, and impossible for the average competent person to realize how little they know about China 2) this contains valuable open-source intelligence and it has helped me out.

China is emerging as the preeminent world superpower, and will likely be a larger economic powerhouse than the US within our lifetimes.

  1. This is false, or rather, it's a bad induction. It's actually a massive toss-up as to whether China will rise successfully, American experts tend to lean towards Chinese failure to rise but there might be systematic anti-china bias there. It's not that biased though, since the US is still the reigning superpower so we can assume the US probably still has many secret capabilities not yet revealed. One way or another, whenever someone says something like this they're basically calling themselves out as not having spent sufficient time researching the topic. In this case, this distracts from OP's open source research skill.
  2. "War" as it's referred to in this post has already been taken into account for biorisk, AI risk, and nuclear risk. It's very unclear whether war would ever take the form of WW2-style war  war. That was the norm for 1930s tech (1940s tech saw the rise of ballistic missiles and  very small nuclear arsenals, both emerged in the early-mid 40s after most of WW2 was already over). The Cold War saw militaries transition to using spies to try to infiltrate and kill off hostile foreign governments. That was as threatening as large land armies, but didn't steamroll over lots of towns and cities the way WW2 did.
  3. These are very sensitive topics in general and it's best to talk about them with people working in the area, not drawing maximum attention to them via forum posts, which distribute the info randomly and to as many people as possible. Personally, I'd suggest a post like "some valuable information and research about the current state of the US-China tech situation" with minimal or no policy prescriptions, which would 1) be very helpful to the people working in this area (as you've already demonstrated) and 2) be more likely to get you yourself connected and integrated into those networks,  allowing you to contribute even more to complicated sensitive topics in a non-risky way
  1. Could you explain why you think China will not rise successfully, without deferring to experts (unless these experts have some kind of testable prediction record)? The Chinese government has managed to bring more people out of poverty in the last 50 years than any other government : the pace of economic growth is staggering. Chinese GDP per capita was $100/person in 1970: today it stands at $10k/person. Profit is not a dirty word in China, whereas growth and profit have become political in the US (i.e, strong cultural headwinds downstream of economic growth in China). China's population is 3 times larger than that of the United States. The economy expands to more and more industries versus being simply a low cost of labor manufacturing destination like the US wants to paint it, resulting in greater technological capabilities. For example, IMO, the most exciting new industrial robotic arm companies are predominantly Chinese. China's geopolitical influence is growing faster than the US's, because China is strategic about its geopolitical interests, planning and executing over 10-20 year horizons: whereas the US seems to be constantly shuffling its State Department and other organs, executing upon 2-3 year long plans at best. These are most of the relevant factors I was using to make the claim: but of course it's very hard to predict geopolitical events. At minimum, we should take the possibility seriously.
  2. Weapons technology may incorporate biology, AI, and nuclear, but judging by the history of military technology during WW1 and WW2, it's quite difficult to predict what the most cost effective killing machines will look like. My bet is on cheap killer robots rather than bio/AI/nuclear. Killer robots could kill off most remaining humans, for example, because that would be significantly more cost effective and thorough than bioweapons/nuclear weapons. This is not factored into bio/AI/nuclear X-risk. There's a big difference between accidental technological failure, and bringing governments' resources to bear on killing the maximal number of people: it's very hard to predict how good governments will be at that, because they are live players and can push technology forward very quickly (and historically have done that). For example, did you know that most engineering research at MIT during WW2 was towards the War Effort? Imagine if most smart engineers were pushed towards weapons research. That level of resources has been missing for 50+ years, and I think it's very hard to predict how bad things could get - but my guess is they could get very, very bad.
  3. Thanks for the feedback, and sorry for contributing to epistemic pollution around the topic. Upon reading this again, I agree it's not nuanced enough, but still stand by my core argument: US elite culture is shifting towards a confrontational lens on China, and away from a collaborative lens, and that if this trend continues, it has huge implications for humanity in the next 20-30 years. Separately, I'm not persuaded by most infohazard arguments, so I have a hard time understanding your perspective on why this would be risky/sensitive to discuss in public: could you please explain?

I'm not persuaded by most infohazard arguments, so I have a hard time understanding your perspective on why this would be risky/sensitive to discuss in public: could you please explain?

The brief answer is that it could cause harm to EA's efforts, to EA people in China, or to EA people working on China-related issues.

The Chinese government is very sensitive. It takes perceived slights quite seriously, and has lashed out in the past against national governments, corporations, and individuals for perceived slights. Any discussion of China that is connected to EA and is publicly available could easily get EA banned in China if it is perceived as against China at some point in the future. That would cause serious difficulties for some EA causes, as well as for some people connected to EA who are working on related issues.

Anything related to China publicly should be done with great caution, as it would be far too easy to get an organization or a movement declared illegal in China.

When one doesn't live in and think about autocratic governments, then it is easy to forget how flimsy of a justification the state needs for putting someone under house arrest, or for taking someone to a detention center.

However, if the research is valuable enough (like the sheer quantity of open source research here, that virtually noone would already know everything regardless of expertise), and if it's private enough e.g. communicated through DMs and in-person, then the benefits usually exceed the costs. China is pretty well known to crack down on outspoken critics, not people who "know too much".

Policy prescriptions, embarrasing/disturbing info about governments, and outright denunciations of foreign governments are definitely going to ramp up those risks. There's also complicated delicate diplomatic situations/negotiations to consider; pareto solutions often involve some amount of facilitating internal situations in other countries since those are the top priority there.