When we say Artificial Intelligence we first think of robots. But AI is more generic than that. AI incorporates the machines, that we use in our day to day lives, with intelligence. Intelligence in the sense that machines do not have to depend totally on humans for all the petty decisions and acquire some sort of independent thinking.
The term Artificial Intelligence was coined in 1957 at a conference at Dartmouth, New Hampshire convened to discuss the possibilities of simulating human intelligence and thinking in computers. Even until now there is no clear definition for AI, but it is a well established, natural and scaling branch of computer science.
The three main category of problems AI deals with are mundane, formal and expert problems. Recently it was proven that mundane tasks are the harder to simulate because they require common sense and hence lots of knowledge. On the other hand expert tasks are domain and knowledge specific hence easier to accomplish for the machines. The solutions for these AI and some other non-AI problems are got using some AI techniques such as Search, Use of knowledge and Abstraction.
The basic issues relating to AI technically are Knowledge representation, Reasoning Techniques and Learning.
Some very famous knowledge representation techniques are: Predicate Logic, Weak & Strong slot and filler structures, Semantic Nets, Frames, Scripts
Knowledge is basically large chunks of data the machine is given initially to acquire the basic intelligence levels. As the machine works, it must be able to acquire intelligence from experience (inference), learn from mistakes and also store these “Life lessons” as knowledge so that it can reuse this experience later when required.
e.g. The following sentences are stored randomly in the knowledge base
1) Neil Armstrong was the first man on the Moon.
2) All men are Mortal.
When represented in predicate logic they look like:
1) First-on (Neil Armstrong, Moon)
2) ¥ x: Men(x)→Mortal(x)
Various Heuristic search techniques such as Generate-and-test, Hill Climbing, Best-First Search, Problem reduction, Constraint satisfaction, Means-ends analysis are used to search for the specific information needed in the knowledge database that uses any of the representation techniques.
e.g. when given the following statements:
1) Neil Armstrong was the first man on the Moon.
2) All men are Mortal.
Now if we ask : Is Neil Armsrtong a mortal?
The machine should reason and answer “Yes” using reasoning because it is assumed that it does not have a specific answer to the above question already in the knowledge base.
Many reasoning techniques are used in the contemporary world like: Formal reasoning, Procedural reasoning, Reasoning by analog, Generalization and Abstraction, Meta-level reasoning, uncertain reasoning and Non-monotonic Reasoning. One of the very often used reasoning technique is default reasoning which is a form of Non-monotonic reasoning.
Reasoning makes use of many mathematical concepts such as Probability, Baye’s theorem and Certainity factors. Various search techniques
Defining Learning technically “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P ,if its performance at tasks T, as measured by P, improves with experience E”
Various learning techniques are: Rote learning, Learning by taking advice, Learning by problem solving, Learning from examples, Decision trees, Reinforcement learning.
Learning tasks covers a wide range of phenomena: Skill refinement, Knowledge Acquisition, Taking Advice, Problem solving, Discovery etc…Computers (like humans) require Vision, Image Processing, Speech Recognition, Sensory motors, Natural language understanding to learn. The adequate hardware and software for this must be provided.
e.g.: A robot in order to move from one room to another requires vision to see, image processing to locate the door and Sensory motor co-ordination to walk from one room to another.
Game playing is a field which demonstrates several aspects of intelligence, particularly the ability to plan (at both immediate tactical-level and long term strategic level) and the ability to learn. e.g.: Perhaps the best example of usage of intelligence in game playing is the Chess Grandmaster supercomputer DEEP BLUE which beat the then present Chess grandmaster Gary Kasparov.
3. Medical improvements:
Computers that take care of human health from both inside and outside the human body have already being developed and are constantly evolving. These computers check the human condition, diagnose it and even suggest a prescription. e.g.. :MYCIN
4. Many more:
Computers that detect molecular structures e.g.: DENDRALComputers that help in editing, space programs, real life graphics,Military simulations, Neural Networks and Fuzzy Logic, Natural language processing, Image processing, Computer vision, Speech Understanding.
On the flip side of the coin these AI machines are becoming increasingly helpful in many crucial areas where man always felt a bit left out without any advice or help. AI machines share the workload of humans both in physical and mental aspects. They lower the work of human experts. AI machines try to bridge the gap between a man and his machine. There is hardly any area where AI is not applicable. AI is fast picking up its pace into the future.
Either way AI perhaps is the most researched and most awaited computer generation and the trend, for the demand and applications for intelligence in machines for human luxury and need, is here to stay.
The term Artificial Intelligence was coined in 1957 at a conference at Dartmouth, New Hampshire convened to discuss the possibilities of simulating human intelligence and thinking in computers. Even until now there is no clear definition for AI, but it is a well established, natural and scaling branch of computer science.
The three main category of problems AI deals with are mundane, formal and expert problems. Recently it was proven that mundane tasks are the harder to simulate because they require common sense and hence lots of knowledge. On the other hand expert tasks are domain and knowledge specific hence easier to accomplish for the machines. The solutions for these AI and some other non-AI problems are got using some AI techniques such as Search, Use of knowledge and Abstraction.
The basic issues relating to AI technically are Knowledge representation, Reasoning Techniques and Learning.
Knowledge Presentation
Intelligence in machines is said to be drawn from knowledge. It is knowledge that is to be acquired, stored and used by the machine in order to act intelligent and solve real world problems. This knowledge is in huge amounts and necessary structures are required to store, search and represent this knowledge.Some very famous knowledge representation techniques are: Predicate Logic, Weak & Strong slot and filler structures, Semantic Nets, Frames, Scripts
Knowledge is basically large chunks of data the machine is given initially to acquire the basic intelligence levels. As the machine works, it must be able to acquire intelligence from experience (inference), learn from mistakes and also store these “Life lessons” as knowledge so that it can reuse this experience later when required.
e.g. The following sentences are stored randomly in the knowledge base
1) Neil Armstrong was the first man on the Moon.
2) All men are Mortal.
When represented in predicate logic they look like:
1) First-on (Neil Armstrong, Moon)
2) ¥ x: Men(x)→Mortal(x)
Various Heuristic search techniques such as Generate-and-test, Hill Climbing, Best-First Search, Problem reduction, Constraint satisfaction, Means-ends analysis are used to search for the specific information needed in the knowledge database that uses any of the representation techniques.
Reasoning Techniques
Reasoning is required to drawing inferences from what it already known. It is required to do something that a machine or any knowledge system has not been explicitly told to do. In this case the machine should reason from all the information that it has in form of knowledge and try to do a task for the first time thinking all by itself. Truly the first step in displaying intelligence.e.g. when given the following statements:
1) Neil Armstrong was the first man on the Moon.
2) All men are Mortal.
Now if we ask : Is Neil Armsrtong a mortal?
The machine should reason and answer “Yes” using reasoning because it is assumed that it does not have a specific answer to the above question already in the knowledge base.
Many reasoning techniques are used in the contemporary world like: Formal reasoning, Procedural reasoning, Reasoning by analog, Generalization and Abstraction, Meta-level reasoning, uncertain reasoning and Non-monotonic Reasoning. One of the very often used reasoning technique is default reasoning which is a form of Non-monotonic reasoning.
Reasoning makes use of many mathematical concepts such as Probability, Baye’s theorem and Certainity factors. Various search techniques
Learning
Any machine cannot be called intelligent until it is able to learn to do new things and to adopt to new situations. Machines learn and store this information for future references. But the main issue is these machines must be provided with adequate mechanisms to learn new things from old. The process of learning produces and increases knowledge and improves behavior and performance of a machine.Defining Learning technically “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P ,if its performance at tasks T, as measured by P, improves with experience E”
Various learning techniques are: Rote learning, Learning by taking advice, Learning by problem solving, Learning from examples, Decision trees, Reinforcement learning.
Learning tasks covers a wide range of phenomena: Skill refinement, Knowledge Acquisition, Taking Advice, Problem solving, Discovery etc…Computers (like humans) require Vision, Image Processing, Speech Recognition, Sensory motors, Natural language understanding to learn. The adequate hardware and software for this must be provided.
e.g.: A robot in order to move from one room to another requires vision to see, image processing to locate the door and Sensory motor co-ordination to walk from one room to another.
Applications of AI
1. Game playing:Game playing is a field which demonstrates several aspects of intelligence, particularly the ability to plan (at both immediate tactical-level and long term strategic level) and the ability to learn. e.g.: Perhaps the best example of usage of intelligence in game playing is the Chess Grandmaster supercomputer DEEP BLUE which beat the then present Chess grandmaster Gary Kasparov.
2. Technological advancements:
Present day computers are very fast and on the way to enter the next (i.e.., fifth)generation of computers. AI is the doorway to that new technological world where computers no longer need humans for the minutest details and spare humans for the jobs that really need their attention. The technical burden on humans would be reduced.3. Medical improvements:
Computers that take care of human health from both inside and outside the human body have already being developed and are constantly evolving. These computers check the human condition, diagnose it and even suggest a prescription. e.g.. :MYCIN
4. Many more:
Computers that detect molecular structures e.g.: DENDRALComputers that help in editing, space programs, real life graphics,Military simulations, Neural Networks and Fuzzy Logic, Natural language processing, Image processing, Computer vision, Speech Understanding.
Conclusion
The anticipation of the fifth generation computers whose domain is Artificial Intelligence astound us to contemplate about the future of Mankind and Intelligent Machine-kind working side by side with equal Intelligence. We should not ignore the future possibilities that these machines could start thinking for themselves and not under human control, of course this is a bit far fetched but cannot be ruled out completely.On the flip side of the coin these AI machines are becoming increasingly helpful in many crucial areas where man always felt a bit left out without any advice or help. AI machines share the workload of humans both in physical and mental aspects. They lower the work of human experts. AI machines try to bridge the gap between a man and his machine. There is hardly any area where AI is not applicable. AI is fast picking up its pace into the future.
Either way AI perhaps is the most researched and most awaited computer generation and the trend, for the demand and applications for intelligence in machines for human luxury and need, is here to stay.
References
- “Artificial Intelligence” ---Elaine rich, Kevin Knight.
- “Artificial Intelligence-A modern approach”---Russel norvig.
- “Artificial Intelligence”---Sudhir kumar reddy
An Introduction to Artificial Intelligence
4/
5
Oleh
siva
