If intelligence is your ability to make decisions based on available information, then it should be the easiest task for computers. There is no lack of information, or data as we say it, and no lack of processing power either. Then what makes it so difficult for the computer to develop intelligence?
The basic problem is that the term intelligence itself is not well defined. When I looked up on the web to find out what it means, I was flooded with so many diverse definitions, that I decided to sum it up with my own understanding.
Being a goal-oriented person, the definition I could relate to the most was the one formalized by Shane Legg & Marcus Hutter – “Intelligence measures an agent’s ability to achieve goals in a wide range of environments.” The key here is that the environment will change in an expected as well as unexpected way as we start working.
– Gathering information
– Filtering information based on what could be useful
– Tagging and storing information for future use
– Restructuring based on what we are looking for
– Computing inferences and assigning weightage to them
– Evaluate what-if scenarios based on different inferences and actions
– Execute actions and measure effects
– Provide feedback to the system (retrain) to increase the effectiveness
The ones that are difficult to quantify include
– The ability to take risks
– The ability to evaluate the impact on human life
One extraordinary ability that human intelligence brings in is the ability to learn from other people’s mistakes.
Remember any example of learning from others?