Taipei: This week at the Bloomberg Ideas conference in San Francisco I had the pleasure of meeting a bunch of smart, friendly artificial intelligence (AI) researchers.
Being the Gadfly that I am, I figured it would be a great chance to ask what, to my mind, was an incisive question. Is China or the US better at AI?
Turns out, that’s a dumb question. My interlocutors were kind in their put downs, but my decades of programming in non-verbal linguistics allowed me to interpret their responses.
AI is a vague enough term that any question about it really needs to be more narrowly defined. Right now, what most people really mean is machine learning, which is powered by data. The theory goes that the more data the better, and whoever has the most, wins. But that’s not why my question was dumb.
There’s a general consensus that the Chinese are probably ahead of their US counterparts in terms of data because not only is it a larger country, but the opportunities to capture information are colossal. A January piece by the Wall Street Journal laid out the argument for China gaining momentum with a quote from Microsoft Corp. president Brad Smith noting that in some parts of the world, privacy laws have the potential to constrict AI development or use.
It’s kind of funny (not funny) watching everyone freak out about Cambridge Analytica, Facebook Inc. and the US election when you consider that one billion people have signed up for Tencent Holdings Ltd.’s WeChat messenger service and use it to share intimate messages, buy products, book taxis, make medical appointments and swap news (censored, of course).
Most of those WeChat accounts are linked to an actual, verified ID and in themselves have become a form of identification. They may even soon become the universal basis for credit scores. But that’s still not why my initial question was dumb.
Other components of machine learning include computing power and programming. (For the advanced among you, I’m leaving aside neural networks because it seems that how they work is a mystery even to AI researchers.)
When it comes to computing and programming, my AI experts didn’t seem to have any consensus. Chinese computing capabilities are fast catching up to the US The nation’s semiconductor industry still lags, but so much money is being thrown at the problem that raw grunt power is helping to make up for any shortfall in technological finesse. US programmers, on the other hand, do appear to have an edge, though I couldn’t give you data to back up this claim. Again, that’s not why my question was dumb.
One way to measure superiority is to track patent filings. Analysis of keywords indicates that China trumps the US in the more general “artificial intelligence" patents, according to European Patent Office data compiled by CB Insights. But in the specific category of “machine learning," the US is ahead. The WSJ article noted that the US has twice as many AI companies as China and spends a lot more on R&D.
Who’s leading AI is a dumb question because it implies some kind of winner-takes-all scenario as though a Cold War is playing out with a defined end-game. But as one AI researcher told me, it doesn’t matter whether a Chinese or an American makes a specific advance because any new knowledge can be built upon by other scientists.
What matters more is the goal and the implementation of technologies that are being developed. It comes down to who is using AI for the greater good, and who is using it for ill. That’s the better question. Bloomberg Gadfly