Technology and Physical Limits

I appreciate what this presentation has to say about technology running up against physical limits, which can, ultimately, impose too great a cost to justify major new developments in a particular field. Ceglowski argues that the internet, and computing in general, will not change significantly from where it is now -- in much the same way aerospace hasn't changed significantly from where it was in the 1970s.  

Unfortunately, the presentation falls short by discounting the potential for artificial intelligence in the near future.  He dismisses it as so much craziness, without offering any real counterargument.  

Sure, current predictions may be wrong.  But we're talking about more than a few fringe thinkers.  Nick Bostrom's survey of AI experts found:

The median estimate of respondents was for a one in two chance that high- level machine intelligence will be developed around 2040-2050, rising to a nine in ten chance by 2075. Experts expect that systems will move on to superintelligence in less than 30 years thereafter. 

Again, a bunch of these folks could be wrong.  But AI can no longer be written off as just science fiction.

 

How Self-Driving Cars Visualize The World

Chris Umson is the Director of Self-Driving Cars at Google.  In this TED talk, he demonstrates what the world looks like from the perspective of a self-driving car.  My first reaction is to notice how complicated and dynamic even an ordinary driving scenario is.

Since watching this talk, I've become more conscious of how I perceive and process road conditions, both when driving and biking.  I've noticed myself looking farther ahead and slowing down in ambiguous situations -- where it's unclear, based on the available data, what I should do.

 

Umsom's big theme here is that truly self-driving cars are only possible through a sort of quantum leap in design.  Self-driving cars must operate categorically differently from human driven cars.  A corollary to this is that self-driving cars cannot be designed by incrementally improving on the designs of semi-autonomous cars.  

Via Kottke.