Customer Reviews

Brainmakers: How Scientists Moving Beyond Computers Create Rival to Humn Brain

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

 

 

If you've ever read 'I, Robot' by Issac Asimov (or anything like it), this book should be required reading for you. It provides a fascinating and candid view into the history, research, debates and a detailed summary of the current progress in the controversial field of A.I. (artificial intelligence). I thoroughly enjoyed this book, and have shared it's title with many who have also enjoyed it. As an author and sci-fi fanatic, I found it's forthright, non-technical manner both refreshing and easy to read (finished it in one night), and the information provided to be excellent food for thought. Not too technical, this book can easily be understood by just about anyone, and SHOULD be read by anyone who has ever enjoyed reading anything involving robots or computers of any kind (even the average Star Wars fan will like this one). 'Brainmakers' is a must-read which has earned a permanent place in my personal library (which I might add is fairly extensive and somewhat exclusive). If you can find a copy, it's worth whatever price you have to pay for it.

Freedman's book is a useful introduction to artificial
intelligence suitable for the lay reader. More technical
introductions exist in the form of the many good textbooks
on AI.
Brainmaker is a fun read but I do have a few criticisms.
Freedman distinguishes "old time" "good old fashioned AI"
g.o.f.a.i. from what he dubs "nature AI." I believe that he
has simply cobbled together some ideas and that his "nature
AI" does not exist as a coherent project.
Freedman seems to think that gofai was not modeled after
nature. I do not agree. Newell studied how people reasoned,
Boole was building a logic of how people think, and Rosenblatt
had real neural nets in mind. It is also inaccurate to call
gofai a failure. Sure there are lots of things that people
can do that computers can't. But there is also a long and
growing list of what computers can do and people can not. AIs
are good a modus tollens, humans are not. AIs are good at
long chains of reasoning and with negated terms, humans are not.
Computers can handle spaces having many dimensions, humans find
it hard to handle 3. Computers are good at probability and
math, people are not. And the list goes on and on.
Freedman's "nature AI" does not exist as a real AI subfield.
He has simply grouped together a number of new ideas, some good,
some bad. He also spends too much time on biology. If
there is any evidence that intelligence requires "wetwear" such
evidence is not presented in the book.