Original article can be found here (source): Artificial Intelligence on Medium
Dennett mention a dozen hazards (to comfortable thinking), not sure if I agree with all of them, but here I list the ones I consider important:
- Competence without comprehension
- Words striving to reproduce (I would use ideas instead of words)
- Consciousness as a user-illusion
The other nine are there, I am just saying that I thing they are included in these or they are not as relevant, in my opinion. It is a dense book full of ideas, so I am sure I am missing things and probably I’ll re-read it after reading some of the references. I read the book without some clear intention, just because I like the topic and even if I am looking for something I cannot define at this point what it is.
I had a few eureka moments while reading the book, one of them, the one I like the most, was about the path to result usually not being the shortest path:
Let’s say you want to go from A to C, not caring about traversing B, paths from A to B and B to C are optimized independently and use the shortest path. There is a shortest path from A to C! And it is more efficient based on the criteria you care about. The thing is, the work to go from A to B and B to C has already been done, you don’t have to make it, so you start from there and you make small increments toward the more efficient A to C path.
I work crafting software and I like it, so I always end up thinking about how things from other domains relate or can be related to code. There are two interesting consequences of this fact that I can think of:
- Entrepreneurs and problem-solvers usually care about these A-B and B-C paths, they don’t usually say, I will make the most efficient way to do sell books online (just an example), they think, I have a group of book producers (providers) and a group of book readers (consumers) let’s see the most efficient way to match this two, that is, the effortless solution. The beauty of this is that it is the way nature works, you cannot afford to start from the ground up every time, evolution take what the previous generation had to solve the challenges of the new generations, and that is not always the most efficient, but is the only way to go.
- Obviously my first thought was dynamic programming: memoization and backtracking. Suddenly I realize how it worked in a very visual way (also it remind me of a documentary on Velcro I saw probably 15 or 20 years ago). Maybe the problem with these techniques is that we try to make them as exhaustive as possible instead of focusing on a criteria that minimize effort and maximize efficiency. In my opinion, the most interesting question is: how can we put paths into a space in which we can effectively measure distance? that is, how can we measure how far away we are from the optimal path and if it is worth getting there?
Competence without comprehension
I have read some reviews and comments about this. I won’t try to approach from a philosophical perspective. In my opinion the point is very clear, it is the way nature works: you put a constraint that make the top-performers reproduce, they don’t need to be aware of what they are doing right, whatever they do makes them survive so they just keep doing it. I love how it can be applied to so many thinks like free markets, google analytics, and conscience and human beings in general.
Words striving to reproduce
I have never thought about the idea of an autonomous reproduction of ideas, not explicitly at least. Pedro Domingos mentions how we learn by either evolution, experience or culture, the he talks about computers, I am not sure why it should be considered something different since the interesting thing is how culture works a way of remote transmission of experience. I think the meme concept, as coined by Richard Dawkins and used by Dennett, is a very provocative idea worth exploring, maybe just for writing a short story, maybe something more scientific. When I first heard the idea of human beings and, more generally living creatures, as just an envelope created for efficient gene reproduction, I thought it was amazing because of its simplicity and elegance. Now, with memes, I think it is another layer of abstraction built on top of that first one, culture and memes make possible the remote transmission of information with the small variability needed for evolution to take place.
Consciousness as a user-illusion
Maybe this is the one I have more problems to explain, I think it is related to competence without comprehension but taking things to a whole new level. Perhaps an answer in this domain was one of the main reasons to make me read this book. I think most things in life are subjective, but how far can we take this claim, well, I think we will discover this soon as the possibilities for creating real-artificial intelligence come closer. For me it is a two ways street, we create artificial intelligence in a biological sense, then we can take biological intelligence to an artificial sense.