Re: AI timeframes

From: J. Andrew Rogers (
Date: Sat Apr 10 2004 - 00:06:50 MDT

On Apr 9, 2004, at 9:15 PM, Thomas Buckner wrote:
> A likely 'Sputnik' source may be China...
> "But what are China's chances of actually outpacing
> Bill Gates and America's other techno-leaders in the
> sprint toward next-generation wonders? Very good
> indeed. (...) Not to mention artificial intelligence -
> where one investor feels Guangzhou's Hua Ling Group
> 'is a step ahead of foreign scientists.'"

China is a very interesting case, but I think it unlikely for two
reasons. (And I discount the opinion of an investor as to what
constitutes "good AI" prima facie.)

First, even though China has a large population at its disposal, it is
extremely under-utilized from the standpoint of being a technological
competitor such that it has, at best, the effective technological
development footprint of something like a medium-sized European
country. And it will be at least two decades before they really get
past this. Very little of China can be categorized as "modern" in any
normal sense of the word e.g. the tap water isn't safe to drink in the
best parts of Beijing and donkey carts abound in much of the country.

Second, while the Chinese do come up with some interesting ideas, they
are famously poor at execution. I don't see this changing any time
soon, as the problems with execution seem to be largely structural and

The important issue that is covered in these two points is that by the
time China could hope to really become a major player, I think most of
us fully expect the game to already be over. Japan easily stands to be
a more important player in any timeframe that matters.

> They're making some first-rate electronics over there
> (I have a couple of Chinese-made Behringer audio
> devices, a small mixer and a digital guitar effects
> box. They cost little and do things that used to
> require a cart load of pedals & such).

Behringer. The same Behringer that has been sued by just about
everyone in the audio electronics industry for blatantly stealing
designs and cloning other companies' circuit boards. The Behringer
that is synonymous with third-rate quality in the audio electronics

Behringer is *not* what you want to use as a shining example of Chinese
quality or ingenuity. But it does hint at another point: It is very
slow and difficult to develop a world-class engineering culture, there
are no shortcuts. Because of the likely overall nature and complexity
of engineering a practical AGI, I find it highly improbable that it
will be successfully developed in a country that does not have a
pervasive native engineering culture (US, Japan, and a small subset of
Europe). It could happen elsewhere, but the odds do not favor it.
You need a broad, extremely capable, and highly experienced engineering
team to have a good shot in implementation, which you can't just throw
together in most parts of the world.

Another tantalizing idea: Someone may have already solved AI, but did
not have the engineering competence to realize it well enough to prove
the concept. The algorithm space is very unforgiving of mediocre

j. andrew rogers

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:46 MDT