Possibility of AI – An Overview

Possibility of AI – An Overview

“The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.” (by Ada Augusta)

When we talk about AI so two groups do contradict with each other. One of them believe that everything is calculated and working on the basis of instructions given to the computer. Others believe that a computer can be same like a human, such people like AI movies and working on these concepts to be in fantasy if i am not wrong.

But an interesting part of this contradiction is that, there is another group which don’t agree with first one and also not with second one. I will call them fuzzy people, those lie in fuzzy group who are not exactly denying and not exactly accepting but partially accepting. They believe that a computer can be intelligent as much as we could cover those aspects which can shape it as an intelligent system. If we will be successful in this effort so a computer can think, no doubt like a human because a human also compare the things and then evaluate that next time it should do that or not?

• the goal of machine learning is to build computer systems that can learn from their experience and adapt to their environments
• computers can not think because they only do what their programmers tell them to do and the same inputs and conditions it will always produce the same outputs
• three phases in machine learning
  1. Training
  2. Validation
  3. Application
• Learning techniques
  1. Rote learning – is no prior knowledge
  2. Deductive learning – works on existing facts and knowledge and deduces new knowledge from the old
  3. Inductive learning – takes examples and generalizes rather than starting with existing knowledge
     • concept learning
• branches of problems
  1. Tractable
     1. structured
     2. Complex
  2. Intractable
• Entropy

  In order to define information gain precisely, we begin by defining a measure commonly used in statistics and information theory, called entropy, which characterizes the purity/impurity of an arbitrary collection of examples.

If we talk about the experience of a human which is database in case of a computer so a computer can compare the things to take decision and then on the basis of antecedent it proceeds further. The key concept of AI lies in comparison of good/right/option 1 and bad/wrong/option 2, who will tell the computer that what is good and what is bad? Although AI base on fuzzy part but still if we are looking for a human like behavior so humans are sure about good and bad. For example if a human will touch fire so he will not touch it again because it hurts same like that if computer will put its metal or plastic part in that fire and couldn’t set rule about fire so it is not behaving like a human. We all know that robots don’t bother to be hurt but we have to code that if any part of the connected body changing its shape, losing its weight, getting a different color code so robot should take appropriate action. If its because of a touch? so get distance from there. If its last action was a walk in or walk out so do undo that actions can be the possibilities. If nothing was triggered by that robot but environment changed comparatively from last saved smooth time to latest readings about surroundings so what should robot do? Programmer can prioritize the actions which robot should be take e.g.

1. Move, Walk, Run
2. Move, Sit, Stand Up, Bend or whatever according to that situation priority wise

These are just examples but by reading the environment, computer can compare what is being changed and how should it react? Now thats another story, how should it react? Programmer will set the sensitivity level according to the computer’s assembled parts.

This article was written just to discuss the basis on which we can go for implementation of artificial intelligence, i have taken robot’s example because it seems easy to visualize. Your feedback would be appreciated.

Thanks,

Fahad