Customer Reviews

In Our Own Image: Building an Artificial Person

5 star:		(0)
4 star:		(0)
3 star:		(1)
2 star:		(1)
1 star:		(0)

 

 

This is a fairly interesting book about an alternative technology to the traditional, programmed artificial intelligence. Rather than write software that inevitably bogs down both in unforeseen real-world situations and from sheer volume (and unintended catastrophic breakdowns), Caudill argues that neural networks can evolve by experience and be trained, i.e. learn.

The neural net, as I understand it, is a series of circuits that are conditioned to the things ("trained") by repetition of action. So they can change the way they do things and supposedly "learn" as time passes. My roboticist friends explained to me that they make "connections" similar to those that are made in the human brain.

Where they will end up is unpredictable - they are still extremely limited machines and like traditional AI devices cannot function outside of rigid environmental confines - Caudill speculates that they will evolve into human-like complexity of mind. A real mind.

This enters the realm of sci-fi, I hate to say, and even quotes sci-fi for the conclusion. Well, maybe. Maybe not. The human brain is the most complex object in the known universe: they are 100 billion cells or so, and they are interconnected into networks that are the means by wich our minds function. That network is what Caudill expects will form.

Recommended as one point of view. It is well written and interesting.

This is not the book I thought it would be. It seems to jump among several topics that may or may not have much relevance with one another. It also switches from science to science fiction with an uneasy frequency, discussing neural networks then H.A.L. in 2001 or switching from literature to the lab.

The author explores humanity. She asks the right question,
"What does it mean to be human?" and has essays on memory, consciousness, potential, learning and behavior as well as musings on the physical properties of the android. What I found missing were such concepts as good and evil, moral and immoral, right and wrong. If androids become self-aware how will they view themselves in context with other creatures?

This is more an exploratory book of preliminary questions, more a survey of contemporary robotic efforts than an actual look at everyday companions. The future is too hazy, the time too distant, the science too new to do much more than forecast and speculate.