It's not too late as China still has a small ways to catch up. And furthermore, the USA (and the UK) are home to the companies paving the most promising paths towards artificial general intelligence.
The infrastructure for a turnaround is still there. If the USA seriously wanted, we would have legions of youth STEMgineers within five years. But that's not what our zeitgeist is about.
In terms of funding basic science, the USA is still the leader, but only by being grandfathered in and still having that infrastructure to work with. I'd trust NASA over the ESA or CNSA— but by increasingly less each year.
And then there's the issue of AI and automation. This is what's going to determine the scientific leader of the 21st century, and it's absolutely true that the USA has dropped the ball with it. The private sector is the one funding AI advancement, and it's seen rapid gains. However, it's not seeing rapid gains because of the private sector— after all, much of the advanced stuff we've been hyping ourselves over for the past decade has been almost completely laboratory work, with the actually implemented stuff being far lesser in comparison. That could've been accomplished with government funding all the same. Having private funds is more of a boost; government funding can be fickle depending on who's in charge; private funding is only fickle if your parent company is in dire straits, and corporations like Alphabet and Microsoft are not sweating.
If the USA were run competently, it'd be a softcore capitalist technocracy, with officials deferring to these industries for any and all opinions beyond their range. But as we see with environmental science, that's not the case. Lawyers and lobbyists believe themselves to be more educated on scientific subjects than actual experts. And those that don't believe it do believe that their constituents are gullible enough to believe they do and don't need to understand any of these more esoteric subjects.
Indeed, the Trump administration barely even acknowledges automation as a threat. I distinctly remember saying back in 2016 that the reason why Trump's plan to restore American industry was going to fail because of rapidly expanding automation, but he and his administration were going to completely downplay or ignore the threat and instead focus on immigrants and foreigners as the cause of MAGA failing. Literally a few months after his inauguration, Steven Mnuchin said that automation was “50 to 100 more years” away.
And Trump has not mentioned AI at all, even since this was published.
If Trump gets a second term (I'm not commenting on the odds as I'm completely unsure; by this point in 2016, I was 100% sure that Trump would win but still assumed Clinton would get the victory; I genuinely can't call it either way for 2020 and humiliate me with my words if I ever say "I knew it" after the fact should Biden or Trump win), this prediction of mine will start really kicking in. It's already been taking effect in limited ways (when you have the time, read reports and articles detailing how some factories now use only a tiny handful of workers while robots do most of the work)— there isn't much potential for the truly sci-fi levels of automation we're expecting because AI is too narrow to handle that, and anything going wrong means the whole system shuts down.
But now take into consideration certain developments that have occurred in the past year, most notably GPT-3.
We literally started the 2020s with one of the most shocking advancements in AI history, the kind of tool researchers from earlier decades could have only dreamed of. It's only going to rapidly get more extreme from there.
Should Trump win in 2020 and somehow makes it to 2024 (because let's assume he's not ousted, doesn't die, and the USA remains solvent through that time), imagine what his deglobalization platform would look like by then. More importantly, imagine what OpenAI's laboratory work will look like by then. If not them, then DeepMind. By 2024, Trump will still talk in terms of immigrants taking jobs, of Chinese and Indians taking jobs, or caravans trying to cross the border to leech off Americans, and that's the chief threat. In the meantime, his economic advisors will still bandy "AI is not a threat for another century, and there is no threat of automation to take Americans' new jobs." They'll be saying this as whatever hybrid transformers and first-gen AGIs are operational at the time are displaying uncannily robust ability to program, design apps, and perhaps even act as the engine for social robots. In fact, if it's 2024, then we ought to see some major advancements in robot control all around, as well as the commercialization of BCIs to both also control robots and feed neural networks with greater data (Starspawn0 has more in-depth knowledge of this and has gone into much greater detail than I could)
The ability to use these neural networks to alter public opinion is endless, as I've also mentioned before.
There's absolutely no chance that China isn't well aware of these things but is just a bit too far behind the USA to realize them themselves. OpenAI isn't a secretive country, and China is home to loads of AI fanatics who can easily access the same sites we can (the Great Firewall isn't as total in its scope as I've heard). They've probably constructed massive transformers but haven't publicly revealed them, and since there's far greater government funding and support of AI there, they really have no impetus to reveal such transformers, especially if they're behind the Western SOTA (the one drawback to government funding is that they don't have to commercialize what they research and thus never have to reveal said research to the public— there are probably black projects from the 1950s that have been rendered obsolete by new black projects that are also now obsolete that we have no idea of). With a technocracy backing them and a comparable amount of wealth to the USA, the only thing they lack is the educated and skilled populace; China's working up from virtually nothing. When the USA was competing with the USSR and creating STEM kids in the '50s and '60s, the Chinese were starving and slaughtering and exiling their own skilled workers and championing the peasantry to become industrial workers via horrific mismanagement and irrational ideological puritanism, neither of which are conducive to creating a skilled workforce or intelligentsia. They were in 1980 where the USA was in 1880. They had to work overtime to catch up to the West (effectively a century ahead of them) in only a generation or two. Why wouldn't they put a massive focus on basic STEM? It's the only way to get there. What are they going to do, focus on calligraphy and painting birds and worshiping flowers and then magically equal and surpass the Western world in 10 years?
The CCP is obviously aware of the potential of AI and robotics, as Xi Jiping hasn't made any effort to hide his interest in the matter. Meanwhile, Trump likely thinks of AI in the same terms as the average American hooligan: Terminator, HAL-9000, and a bunch of socially awkward pock-faced nerds with pocket protectors making robots to impress the popular girls. But to that end, it's a lot like the quiet middle-class-from-poverty Chinese kid working his way up to overcome the rich suntanned white kid who had everything given to him and feels entitled to success. In the end, the white kid might still win, but the Chinese kid's going to give him such a run for his money that he'll be spooked by the experience.