In order to understand this book in context, consider the following analogy. Birds fly, and airplanes fly. But it isn’t the same “fly.” A flock of birds can take off without running into each other, but if a flock of planes tried that, it would be a disaster. On the other hand, an airplane can fly across the ocean, while few birds can make that trip. We use the word “fly” metaphorically, but the metaphor is far from perfect.
So, when we use the same word “intelligence” to describe something humans do and something machines do, we risk making the same mistake. There are things machines can do which humans cannot (like the birds trying to fly across the ocean in the analogy). For example, machines can crunch incomprehensibly large datasets and reduce them to one, of possibly many, comprehensible summaries. Similarly, there are things humans can do which machines cannot (like the flock of birds taking off without colliding with each other in the analogy). For example, humans can bond with each other, enjoy social interaction, prefer one item on a menu over another, or plan a self-interested future.
The author of this book seems to equate artificial intelligence (AI) with machine learning, which is not only incorrect but dangerous. It is dangerous because there are things that machine learning can do which humans cannot, such as make better specific decisions after having gone through vast amounts of data. We worry about this because we fear machines taking jobs away from people, which is certainly a valid concern. But the real worries are those which we currently cannot even imagine. The author is concerned with potential outcomes and the fact that too few people understand the technology or the outcomes. Both are valid concerns, but should be considered in the correct context. Nonetheless, addressing the concern that too few people understand the technology or the outcomes, the bulk of the book provides a history of advances in human understanding that led to where we are today.
The justification for the historical approach is that if one understands the evolution of human intelligence, one can understand more about the evolution of machine intelligence. The history begins with the Big Bang, which is a little odd because, if the universe were a day old, humans did not show up until a minute before midnight on New Years Eve, and what we think of as human intelligence did not show up until seconds before midnight. Machine intelligence showed up, in turn, a tiny fraction of a second before midnight. But, perhaps, that’s a little nitpicky. Still, it provides a perspective on the problem.
The book goes on to explain the emergence of numbering systems and mathematics, which are more clearly related to machine learning. Then to early computers as code breakers, followed by the rise of programming languages, video games, and eventually big data. The history is interesting, but it does get a little tedious at times. Following the historical survey is a section on what world powers such as the US, China, and Russia are doing, and how emergent AI is an influence in global power. This is an important section; what people will do with AI is a greater concern than what AI will do to people. Again, it is at times interesting and at times tedious.
What the book lacks is an overarching theme that ties together the advances in human knowledge in such a way that we can apply it to the trajectory of AI. There are many possible overarching themes; since the author does not provide one, I will. The history of the emergence of human intelligence can be seen as humans being human (that is, curious) and discovering new things in the process. Those new things almost always had unintended consequences. Some of them were massive. For example, when language was emerging, people were making noises at each other in a primitive attempt to communicate. Nobody said:
Let’s create a language so that one day we can create an alphabet to write it down. Using that written language, we can record human knowledge across generations so that knowledge is not limited to what your parents can teach you in the first couple years of life. Eventually, if we stay focused, we will have Wikipedia and chat rooms.
No, nobody ever said that. It is more likely that some hunter with no name made a guttural noise at another hunter with no name, suggesting that the second nameless hunter attack the mammoth from the back while the first hunter attacked from the front. Knowledge encoded in written language, Wikipedia, and chat rooms are just unintended consequences. I make this point because the author wants people to understand AI so that the democratic process can guide development. Good luck with that!
Returning to the beginning of the book for a moment, the author begins with five questions that need to be addressed:
- (1) What exactly is AI?
- (2) What aspects of our lives will be changed by it?
- (3) Which of those changes will be beneficial and which of them harmful?
- 4) Where do the nations of the world stand in relation to one another, especially China and Russia?
- 5) And, what can we do to ensure that AI is only used in legal, moral, and ethical ways?
Only the fourth question is specifically addressed. It appears that the author’s answer to the first question is that AI is machine learning. However, over the years, AI has been diagnostic systems, image recognition, speech recognition, and various other attempts to use technology in specific ways to improve (or sometimes mimic) human performance. However, with an umbrella definition that broad, it would be hard to find any technology that would not qualify as AI. The second, third, and fifth questions are unanswerable. If you replace AI in each question with “unintended consequences,” the reason they are unanswerable becomes obvious.
More reviews about this item: Amazon