Intelligence, Infallibility and Creativity.

I've been reading Roger Penrose's book "Shadows of the Mind". The book makes rather interesting reading and makes one think a lot about a bunch of things that most of us don't think about much.

The basic point the book tries to make it that the human brain is more than a computing machine. One quote from Turing that caught my attention was
If a machine is expected to be infallible, it cannot also be intelligent. There are several mathematical theorems which say almost exactly that. But these theorems say nothing about how much intelligence may be displayed if a machine makes no pretence at infallibility.

The quote somehow suggests that just a little bit of "fallibility" would go a long way in making machines intelligent and possibly suggests that allowing for a few mistakes will allow machines to reach the level of human intelligence. Clearly computers are better than humans at performing tasks using known procedures. I suspect that "creativity" or the ability to generate "new" ideas would be considered an important attribute of "Humans" that the machine would be expected to match.

A honest look at myself suggests that the machine would be required to be very significantly fallible. I find that a lot of the "ideas" that I come up with do not actually work well. I would conservatively state that about one in ten of my ideas do end up surviving serious scrutiny. I also know quite a few people (who I consider intelligent) who have at various times indicated that the ratio of "correct" to "incorrect" ideas they have is quite low and in the order of one in ten.

In my own work on successive approximation ADCs, I've found many instances where allowing circuits to make errors results in the system working better than trying to make the circuit "correct" at all times. The idea of error correction and redundancy are quite commonly used in many circuits to improve the performance of the circuit itself. My experience with such circuits suggests that one needs to allow significant errors before the advantages become significant.

In the biological sciences one encounters the phenomenon of evolution as a case where the fallibility of the DNA replication process results in the origin of new species over time. DNA replication is a very accurate process which is probably why evolution takes such a long time. It is also quite possible that the "randomness" of the process also contributes to the slow pace at which new species are "created". This could also suggest that human creativity is not entirely a random process. It can also be argued that "creativity" is not deterministic either since it is not clear what determines the "creation". The even more bizarre phenomenon of intuition probably plays a bigger part in human creativity.

One question to the readers of this blog is "What fraction of the new ideas you get end up being valid once you put them to serious scrutiny?" I would appreciate if you could leave that answer as a comment.

I also found myself wondering if "creativity" is indeed a result of intelligence. Again a simple first person introspection throws up an alternative view. Lets say I want to solve a new problem that I've encountered. My intelligence tries to solve the problem using all the tricks it already knows. Some of these tricks are useful and make some headway towards finding the solution. A lot of these tricks are however not useful and don't contribute much. A lot of effort goes into solving the problem after which a "flash of intuition" occurs which presents the "creative" solution. Another mechanism that generates the "flash of intuition" is listening to some one else describing how they have solved the problem. One goes about hearing and understanding the other person and suddenly a new idea emerges. I don't remember a time when I got a brilliant idea that solves a problem that I've not spent a lot of time thinking about.

I'm wondering if creativity occurs only when intelligence is put out of the way. In other words, creativity occurs only intelligence has been "satisfied" and/or "exhausted". One might even be tempted to argue that intelligence has been a hurdle to overcome before the creative process kicks in. There are many examples one hears where experts in a field fail to see a simple solution while a lay man who just happens to walk past offers a shockingly better solution. I know a lot of people who point out that they have been working in the same area for too long and find themselves unable to be as innovative as they had been in the past (I'm saturated feeling). It is quite possible that such "experts" will need to exhaust a large bag of tricks they already possess before they can think of a new solution to the problem.

It would be interesting to think of ways to "satisfy" one's intelligence as a way to improve ones creativity. Clearly working very hard on a problem is one way. Reading or listening to others who have solved similar problems is another. A better alternative would be brainstorming where a group of people discuss the problem in detail. Usually a lot of ideas are discussed in such sessions and quite often people do end up getting some bright ideas in the end.