Powered by glolg
Display Preferences Most Recent Entries Chatterbox Blog Links Site Statistics Category Tags About Me, Myself and Gilbert XML RSS Feed
Sunday, Dec 01, 2019 - 23:51 SGT
Posted By: Gilbert

Tech Of Our Generation

So, here it is - the new new hype, or the new old hype depending on how long you've been around: artificial intelligence. The state's announced its half-billion dollar National AI strategy plan about a fortnight ago, on top of the S$150 million a couple of years back... pretty much like every other country out there, it has to be said. It might not have been much of a choice. My personal theory is that modern world history has revolved around ascendacy in what might be termed as "generational technologies":
  • ~1910 to ~1950: the golden age of (nuclear) physics, best characterized by the Manhattan Project; changed geopolitical strategy evermore with the mutually assured destruction doctrine, investing nuclear weapon owners an outsized deterrent effect
  • ~1950 to ~1970s: the Space Race contested largely between the two great (nuclear) powers of the post-WW2 era, America and the Soviet Union. The early U.S.S.R. lead was swiftly lost, and the U.S.A. achieved symbolic victory with the first moon landing in 1969
  • ~1970s to ~2000: digital consumer technology and the Internet. The Soviet Union's decline and dissolution left the American-led First World in undisputed control, and perhaps the true flowering of globalization, with the world no longer divided into ideological blocs. The dot-com bust dampened enthusiasm somewhat, but as we have seen, it was more of a temporary hiccup
  • ~1990 to ~2000s: a life sciences boom overlapped the above, with its showpiece achievements probably animal cloning and the Human Genome Project; interest has arguably waned for some years, however, with the discovery of many fundamental obstacles to the practical application of biotechnology
  • ~2010s to ???: the third major Artificial Intelligence summer, after the winters of the late 1970s and early 1990s, leading to dormancy through most of the 2000s

Note that terming these as "generational technologies" does not imply that there were no significant developments in other fields, or that no progress happened in those areas after the time period stated - nuclear fusion is still being worked on, SpaceX and satellites are still a thing, as are various shiny web stacks and groundbreaking bio tools like CRISPR. It was just that those technologies more or less defined their generation; the best and brightest of the 1930s would arguably have been nudged towards physics, of the 1950s aerospace, and most recently, biology and perhaps quantum physics (computer science in the 1990s was perhaps something of a special case, since formal instruction was not strictly required to excel there). Seen in this light, a general consensus appears to be forming on A.I. being the thing to do, for the foreseeable future.

It can also unavoidably be observed that mastery of the generational technology of the era has been strongly correlated with global influence and power. The Cold War was led by those with the most nukes, and the U.S.S.R. clearly fell away after being outdone in space. The rise of the Internet coincided with the near-unfettered spread of Western (i.e. mostly American) liberalism, and leadership in biotech has kept the U.S. and Europe dominant in pharma and its accessory medical sciences.


The China Challenge

The U.S.'s streak of four generational technologies in a row might however be under threat, with China apparently pulling out all the stops on A.I. in particular, having already possibly edged ahead in 5G. Their Big Three of Alibaba, Baidu and Tencent have just been added to an official "A.I. national team" (because, let's face it, can they refuse like, say, Google would?)

America, seemingly caught flatfooted, has responded by banhammering Huawei's 5G products, and is now going after A.I. startups from China, alongside hollering up supposed academic and industrial espionage. In particular, there has been pushback against the strategy of using Chinese nationals embedded in American labs to divert funding and effort towards boosting China's own technological base. About this, it should first be said that the prevalence of Chinese nationals in STEM had largely been an economic quid pro quo (see: case of ResNet): U.S. universities and institutions got brainpower and work ethic on the cheap, while the students received cutting-edge mentoring and facilities, and a fat paycheque by China's standards. A decoupling of this symbiotic relationship, analogous to the trade war, would be messy to say the least.

But back to Singapore. You've heard of the "transformative" masterplan with five national A.I. projects and the ultimate goal to become a, hubba hubba, global A.I. hub, as always. Please allow me to restore some perspective here, because if you want the self-congratulatory backclapping, there's plenty of that to go around in the mainstream media. To begin with, is some S$700 million pumped into A.I. a lot? Very likely for the common taxpayer, who might understandably ask if they could have just held the GST increase back or something. In the bigger picture, however, it's... honestly not very much. OpenAI alone, for example, has an endowment of at least US$2 billion, and that's just one non-profit. Get to the Alphabets and Tencents of the world, and the bare truth is that the resources that they can and have thrown at A.I. dwarfs that of universities, ours included. Honestly, as with the Space Race, I see only two real players here: China's national team, and America's tech titans (amusingly, a Russian A.I. advised citizens to flee the country)


Our Place In The Brave New World

This is however not to say that we might as well just save the money and give up, though. Returning to the point about supposed Chinese stealing of research - shadowy alphabet agencies aside, isn't the knowledge gained in university, and often national, labs supposed to be freely disseminated anyhow? Heck, publishing work is how the researchers gain credit and promotions! The answer, I believe, is that a crucial part of the accumulated scientific expertise is not openly distributed - oft unintentionally. For instance, wet-lab fields are notorious for having unspoken practices within protocols that complicate replication, and even for subjects such as computer science that in theory allow perfectly-reproducible experiments, there remains much hidden knowledge that accrues with experience. As such, attempting to "free-ride" by working only from published research is unlikely to turn out well.

Our leaders have acknowledged the obvious - that America and China will be driving the A.I. race - and correctly recognized that our best hope is to carve out a niche. The fundamental problem remains as follows: the government wants returns, and quick, what with their foreign adventures generally... not going too well (e.g. the Our Most Successful Investment Firm-led consortium's US$4 billion project in Andhra Pradesh seems to have fallen through, to the usual gloating from netizens, and with Magic Leap patents used as collateral, it don't look good on that end either). The trouble is that good R&D simply doesn't work that way, as a local scientist pointed out on A*Star's refocusing on industrial (i.e. profitable) research last year. In his own words:

"...Knowledge creation is the one part where writing grants make no sense. Grants require a researcher to spell out exactly [what they'll] do in the next 3-5 years, and what milestones they will hit... in exploratory projects where breakthroughs come from unexpected directions - grants are pretty meaningless, and end up restricting researchers to do incremental results which they know will work! In the end, true breakthroughs get stifled."

To this, two responses, one old:

"If you pay a man a salary for doing research, he and you will want to have something to point to at the end of the year to show that the money has not been wasted. In promising work of the highest class, however, results do not come in this regular fashion, in fact years may pass without any tangible result being obtained, and the position of the paid worker would be very embarrassing and he would naturally take to work on a lower, or at any rate a different plane where he could be sure of getting year by year tangible results which would justify his salary. The position is this: You want one kind of research, but, if you pay a man to do it, it will drive him to research of a different kind. The only thing to do is to pay him for doing something else and give him enough leisure to do research for the love of it." (J. J. Thompson)

and one new:

"Maybe it's like the inapplicability of law of large numbers and lack of critical density in SG's scientific community that makes knowledge creation infeasible. It's like mining for bitcoin. Even though, for example, on average you can mine a coin every month. But if you only have one machine it's not strange that you get unlucky and took six months to mine one coin. Application is the key. Singapore's not going to beat Google, Amazon or Tencent in smart city technologies and research. Focus on execution, optimization and commercialization."

[To be continued...]



comments (0) - email - share - print - direct link
trackbacks (0) - trackback url


Next: Island State Endowed With Natural Moat Wants To Build Bridges Not Walls


Related Posts:
A Tale Of Two Countries
Incumbent Investigations
Twelve Mostly Men
Apology And Analysis
Dragon Prince's Dilemma

Back to top




Copyright © 2006-2019 GLYS. All Rights Reserved.