Re: Hans Moravec

hal@finney.org
Fri, 19 Nov 1999 09:40:30 -0800

John Thomas, <jwthom@earthlink.net>, writes:
>
> Apologies if someone's already called attention to this, but the
> Disinformation web site is featuring the work of Hans Moravec at
> http://www.disinfo.com/disinfo?p=folder&title=Hans+Moravec with a number of
> interesting links and interviews.

One funny note: Moravec now predicts, "Machines will attain human levels of intelligence by the year 2040." In his earlier book Mind Children, published in 1988, he predicted that computing power capable of human-level AI "would be available in a $10 million supercomputer before 2010 and in a $1,000 personal computer by 2030." So in the approximately ten years since that book he has pushed out the achievement of AI by ten to thirty years. This is not a very encouraging rate of progress...

BTW Moravec also has an article in the December Scientific American about future progress in robotics.

The article above leads to http://www.frc.ri.cmu.edu/~hpm/book98/ which hold links to commentary relevant to Moravec's Robot book. The links for chapter 1 include the discussion organized by Robin Hanson about the Singularity, while those for chapter 7 include a review of Damien Broderick's book The Spike.

There is a link from chapter 1 to a page of mathematics by Moravec similar to what I presented recently showing how a true Singularity can be predicted to arise mathematically with certain assumptions.

Moravec suggests (W is "world knowledge", V is computer speed):

> Suppose computing power per computer simply grows linearly with
> total world knowledge, but that the number of computers also
> grows the same way, so that the total amount of computational
> power in the world grows as the square of knowledge:
>
> V = W*W

Then he assumes that when just people were involved, dW/dt was constant, but as computers become involved, you need to add a term proportional to V:

> also dW/dt = V+1 as before
>
> This solves to W = tan(t) and V = tan(t)^2,
>
> which has lots of singularities (I like the one at t = pi/2).

The problem with this line of reasoning is that it is unlikely that the rate of growth of knowledge can continue to be proportional to total computing power. Diminishing returns will kick in, and further there will be "friction" losses and inefficiency in coordinating the increasing number of computers (and people) in the world.

Hal