Monday, September 12, 2011

Mind to Intelligent systems

In my opinion, understanding mind is an important activity to develop intelligent systems. The mind is an extraordinarily complex system. Hence understanding mind is complex task. At the same time, the study of mind is exciting and important for web evolution because Mind is the only ultimate source of intelligence. By understanding the mind, we can improve design of web. My analogy is that if Mind is the software, then Knowledge is the content of the software, brain is the hardware and thinking is running software. Thinking can best be understood in terms of representational structures in the mind and computational procedures that operate on those structures. (Analogy derived from the development of computers)  Knowledge in the mind consists of mental representations (Data Structure) People have mental procedures (Algorithms) that operate on mental representations (Data Structure) to produce thought and action, that is, the processes/procedures, applied to the representations, produce the behavior.
The mind has mental representations analogous to data structures, and computational procedures similar to algorithms. Schematically:

Program
Data structures + Algorithms (Mechanical procedure) = running programs

MindMental representations + Computational procedures = thinking

There are six different approaches to representation and computation of Mind
Mind is 
- a logical system
- a rule-based system
- a concept-based system
- an analogy-based system
- an imagery based system,
- and a connectionist system (artificial neural networks)
(I will explore more about these in future posts)

If you have seen the Matrix and Terminator movies, you are aware of the scenario in which intelligent machines eventually come to dominate and mistreat humans. This is not a crazy scenario. Some experts have estimated that increases in computer speed will make human-level intelligence in machines possible within a few decades (Kurzweil 1999; Moravec 1998). We might hope that machine intelligence would treat humans well, but there is no reason to expect that super intelligent computers could or would be programmed to hold paramount the interests of humans. Kurzweil estimates the computing speed of the human brain at around 20 million billion calculations per second, based on 100 billion neurons, each with a thousand connections and the slow firing rate of 200 calculations per second. Assuming continued exponential increase in chip speed, digital computers will reach the 20 million billion (1015) calculations per second mark around
2020.

Daunting task to write the billions or trillions of lines of code that would be needed to enable the computers of the future to approach human cognitive capabilities, but this is possible only if evolutionary algorithms will allow computers to develop their own intelligent software.

No comments: