My rating: 4 of 5 stars
I have never read a book similar to this one! It’s amazing and will make you think and wonder at each step.
The book consists of following two parts:
In the first part the author shows how to build autonomous bots using plausible electrical / mechanical design. These bots resemble to small robotic vehicles. The initial designs are simple and give the basic idea to the reader. And every new vehicle builds on a previous one and adds more features and complexity. The most curious feature of these vehicles is that although they are purely mechanical, their resultant behaviour is similar to human emotions like love, fear, aggression etc. At one point the author even argues that these vehicles can ‘get’ new ideas or even follow a sequence of thought but sometimes it is not very convincing.
Many parts of the book throw light on current day artificial intelligence and machine learning concepts like Turing machine, prediction, classification, concepts similar to gradient descent, correlation vs causation etc.
One of the main lesson of this book is “the law of uphill analysis and downhill invention” which is highlighted multiple times. It means that it is comparatively easy to design a system having some desired characteristics (downhill invention), on the other hand given a system as a black box it is very difficult to come up with its exact design details (uphill analysis). This becomes evident in numerous vehicle designs in the book.
In the second part of the book, the author links back to research in biology which is aligned to the concepts discussed in the first part of the book. This part might be tricky to understand specially for people with non-bio background. Even I just skimmed through this part.
Overall, it’s a very intriguing read! However, it is a bit difficult to understand and might require multiple reads through some sections to get it.
This book also makes me wonder that if these simple designs can lead to such complex behaviour in these vehicles then our brain and associated system is way more complex than than and it is hard to even imagine the kind of resultant complex behaviour it leads to in humans (which it actually does)! Also based on uphill analysis, given a specific human behaviour, it is hard to find how exactly it pans out in the backend system of our body.