All done by incremental improvements.

It’s a form of compound interest. It lacks the panache and excitement of gambling, but in the long run it has profound consequences.

Playing better chess was a problem they wanted to solve, yes, and it has been solved. But there were other goals as well: to develop a program that played chess by thinking like a human, perhaps even by learning the game as a human does. Surely this would be a far more fruitful avenue of investigation than creating, as we are doing, ever-faster algorithms to run on ever-faster hardware.

With respect to Mr Kasparov, he’s missing the point. AI researchers persist with chess merely because it’s a well-understood testbed for incrementally improved searching and pruning algorithms. Those algorithms have extremely broad applications outside of chess, so the continued pursuit of their perfection is meaningful and fruitful.

In a sense AI is doing to romanticism what evolution and physics have done to God. What we consider to be “uniquely human” is continuously being diminished by machines that can do it much better. We’re becoming Minds-of-the-Gaps.

There are still many very difficult problems and drawbacks in AI, but it’s just flat out silly to suppose that the field isn’t making important progress.

And here’s Kasparov’s annotation of the game which is fascinating – shows how, if you’re a human – even Gary K, so much of the game has to be played on intuition.

http://web.comhem.se/talschess/replay/kas-top.html

A fuel efficiency of a megametre per litre is simple to achieve. The Tour d’ France riders achieve 3.5 megametres for 0 litres, they ride 3500 kilometres every year with no fuel at all.

This is of course complete rubbish. Although bicycle riders are high up in the efficiency league tables, never the less they still consume fuel, and still produce carbon dioxide. We will save neither the environment, nor ourselves by believing fairy tales.

If you want anyone to take it even slightly seriously, I’d suggest removing the howler from the first paragraph.

Playing better chess was a problem they wanted to solve, yes, and it has been solved. But there were other goals as well: to develop a program that played chess by thinking like a human, perhaps even by learning the game as a human does.

KnightCap learns chess by playing games, probably not in the same way that a human does. The concept could no doubt be taken further except that chess is considered a bit of a solved problem. All the cool kids are playing poker these days. The search for games where humans can consistently beat computers is itself a kind of interesting little adventure.

There was a neural network backgammon that learned by playing enough games to tune the network into a meaningful statistical model (big number of games). Google knows what the name was, I can’t think of it. Suffice to say that the concept of learning game engines has not been ignored.

Our best minds have gone into financial engineering instead of real engineering, with catastrophic results for both sectors.

Although it doesn’t take a great mind to sit there writing your own paycheques, the question becomes somewhat recursive when you ask who is the bigger bozo — the guy who made the scam, or the guy who let him get away with it. I regularly get stuck on that question :-(

To be fair to the financial engineers, incremental improvements can be highly successful in some engineering applications. The calculating grunt of modern microprocessors is staggeringly more powerful than the original 4 bit microprocessors still within living memory. The real-time 3D rendering that you can do on a $50 graphics card would have been a year of supercomputing power and million dollar budgets a few decades ago. All done by incremental improvements. I suspect that the majority of engineers in most disciplines are quite comfortable with the idea of working within a budget.

By the way, what about Chess Boxing? A perfect example of human ingenuity at its finest, and a far bigger leap into the unknown than any incremental improvement.

… the markets aversion to serious innovation, its tendency to move incrementally towards lower levels of innovation leaving really fundamental and speculative innovation to others.

Naughty market! Someone should give it a good talking to (is it really right to treat the market as an agent?).

I’m constantly amazed by everyday technology. Sometimes I feel like I’m in a sci-fi movie. Microwave ovens, post-it notes, mobile phones, ipods, computers with graphical user interfaces …

Sure some of the innovation has come from governments (eg the internet). But most of it is thanks to people working in private sector companies.

Even so, perhaps it’s true that leaving innovation to others can be a profitable strategy for large corporations. Xerox didn’t do well with their Star computer system. But in time the innovations did very well in the market place. Both Apple and Microsoft ended up using some of the ideas Xerox pioneered (there were law suits over the IP for years).

And who would say that the financial services sector has been lacking in innovation recently?

I think part of the problem with climate change is a sort of technology hubris. I have many friends – highly educated professionals – whose reflex response is that a technological solution will be found shortly and therefore no great sacrifice is necessary today. When I ask them what they base this view on, they usually point to the amazing innovations of the past century. The only problem is the rapid pace of innovation we have since the Industrial Revolution is not the normal state affair but is rather highly unusual in human history. We cannot extrapolate from it.

End result: innovation does need to be encouraged but cannot be relied upon. What do we do if the next two hundred years trun out more like 1000-1200 than 1800-2000?