216
PARTISAN REVIEW
person's lifetime, so people didn't really notice. And even in this century,
people didn't notice, by and large, the exponential growth of computing
until the mid-sixties when Gordon Moore came out with his first version
of Moore's Law and then in the seventies refined it. Even then, it was
only known among a small group of computer scientists. The general
public has caught on
to
this only during the last few years. Exponential
growth starts slowly and is flat and unnoticeable for a long time until it
reaches that explosive knee of the curve.
David S. Rose:
I found today's discussion fascinating.
If
one accepts your
prediction of the development of true artificial intelligence, and combines
it with logically equivalent developments in prosthetic robotics, it is easy
to
envision a time not far off when we will truly be living in the world
that Isaac Asimov created in his "I, Robot" series. However, I followed
your thought process all the way up until the final two minutes of your
presentation. By that point, you had us convinced that super-intelligent
creations (call them robots, or simulacrums, or avatars, or spiritual
machines) will exist with intelligence and, we can also assume, with
physical capabilities far in excess of the humans who created them. But
then in the next breath you dismiss the alarmist idea of "hordes of evil
robots coming over the hill." While I don't necessarily disagree with you,
I also don't see how you can come
to
that conclusion .
It
seems to me that
after a carefully reasoned series of technical projections, you are taking a
leap of optimistic faith that things will somehow just work out for the
good. Is there any way to realistically predict whether the "hordes of
robots" will be evil, benevolent, or completely without character?
Ray
Kurzweil :
That is a good point. I think there are a lot of possible
downside scenarios, and I talk about them in the book. One downside
scenario that is commonly posited is that when you have these non–
biological entities that are much more intelligent than humans, they
are going
to
be quite powerful. What are they going to do with the rest
of us? Maybe they won't see much use for us. Another downside scenario
involves nanotechnology; it will be immensely powerful by the end of the
twenty-first century, and we will be able to create any physical entities
we want with nanobots. But in order for these nanobots to be useful, you
need billions or trillions of them. The only way to create that number is
to have them self-replicate. That's the same solution that nature, or bio–
logical evolution, found . We get from one cell to trillions of cells that
make a human being through self-replication. Well, all of those cells need
to
know when to stop self-replicating; if any of them forget and all the
