What is intelligence and what is not?
The following is my take on the book "On Intelligence" by Jeff Hawkins.
The biggest mistake is the belief that intelligence is defined by intelligent behavior
What is intelligence and what is not? Alan Turing, the British mathematician, believed that behavior and intelligence are tantamount, however, ELIZA has proven it to be a fallacy. Behavior is actually a manifestation of intelligence. John Searle, an American philosopher, took the opposite approach with his Chinese Room experiment. Searle did not define what intelligence is; he stated that computers do not have it.
Are we intelligent enough to create truly intelligent machines? If a machine like IBM’s Deep Blue comes to mind, think again. Many scientists were, are and will continue to embark on this endeavor. Hawkins emphasizes that a truly intelligent machine could be built by examining the schematics of the human brain. Will this approach confine us? Should we limit ourselves to the five human senses? Or, should we explore other sensory inputs like tentacles for tactility and radar for range detection? In fifty years, humankind has advanced from a room sized computer to a pocket size one. With that in mind, when will such machines be readily available? Hawkins argues that with our (and I hope no alien is reading this) advanced way of thinking, it will not take too long.
Our brain is a neural network with the neocortex as the seat of intelligence. The neocortex, in its hierarchical structure, is a memory system that harvests patterns it gets from our sensors. This system is our model of the world and is built from a fraction of what we actually sense. This model is stored in a lightweight and highly optimized version, a sequence of invariant representation. Amazingly enough, we can easily extrapolate the image / melody from seeing / hearing just a fraction of it. This is due to the sophistication of our brain to recognize patterns. According to Vernon B. Mountcastle, an Emeritus of Neuroscience professor, our neocortex ‘runs’ common cortical algorithm in each one of its columns. This algorithm allows us to make predictions efficiently. Hawkins argues, that making predictions is the primary function of our neocortex and is the crux of intelligence. Creativity, which often relates to intelligence, is pattern matching by analogy to past experience from our memory model.
I never thought of our brain in such a way. Obviously, I missed out on a lot. Thank you Jeff Hawkins for this remarkable and enlightening book.
"Hawkins argues, that making predictions is the primary function of our neocortex and is the crux of intelligence."ReplyDelete
I would simply say that the most important ability in humans is the ability to make accurate and timely predictions. To this end memory is vital. Memory is stored in every organ not just the brain. The ability to calculate is an important step in the evolution of intelligence but it is not vital. (I have some thoughts on computational fluency: http://blownfx.blogspot.ca/2014/05/702-moneyball-its-not-just-about-numbers.html
E = MC². First energy, then matter. But why? Some night we will drink too much wine and discuss this. :-)