Charles Stross's Saturns Children and Thoughts on AI

Finished Charles Stross's Saturn's Children this weekend.  It was a great read.  The "sex scenes", which obliquely referenced sex but not directly, are a welcome break from the modern fiction these days.  Given that the heroine is a pleasure robot, the sexual references were exactly right, relevant and integral to the story, without detracting from it.  Compared to, say, Tom Clancy, this book is a textbook example of addressing this subject.  One suspects that the extraneous sex scenes in books these days are a plague inflicted by the book agents.

In any event, Saturn's Children paints a thoughtful picture of Artificial Intelligence inheriting the earth.  The classic tragedy of robots bound by the human property laws, in a world where humans are extinct, is haunting.  The constraints of economic reality on robots' actions is a great reminder that, despite the hopeful communism of Star Trek, et al, Resource Limitation is the universal condition no matter the scale of your civilization.

The notion of Artificial Intelligence having to pattern the human brain to achieve symbolic manipulation, is in line with current thinking on AI.  The spontaneous emergence of consciousness is the combination of the hardwired neural complexity and the profusion of environmental stimuli.  To recreate that neural complexity in software form is probably too expensive, compared to implementing that complexity in hardware form.  Moreover, the role of the environmental stimuli is crucial in guiding the development of the neural complexity.  Without the interaction with the environment, a brain is only latent, not realized.  There are some interesting research taking place on this particular AI hypothesis, and we will soon find out if the hypothesis hold any validity.

One note of caution, though, is that current researchers may be underestimating the amount of environmental stimuli required to realize intelligence.  Brains emerged to interact with the environment, not just to observe it.  Without interaction with the data stream, these new AI brains may starve just like the sensory-deprived orphans of post-communism Russia and Romania.  Or worse, the Hikikomori phenomenon in Japan.  One might think that the current projects are somewhat flawed because they are focusing on building the brain itself, rather than the educational apparatus required to realize the intelligence.

If the prototypes succeed, Dr. David Brin's future of AI working alongside man will be close at hand.  That future of augmented intelligence (where a brain can interact with a digital computer) definitely sounds exciting!