Artificial intelligence is the science and engineering of making mechanical systems that can perform tasks that normally require human intelligence. Such systems may have the capability of visual perception, speech recognition, natural language processing, and they may also be able to make decisions and perform tasks without human intervention. Artificial intelligence is generally implemented as computer software that may be used to operate robotic systems.
People have been fascinated from ancient times by the idea of using mechanical devices to perform complicated tasks. Around 200 BC, the ancient Greeks developed the Antikythera mechanism which was an analog computer designed to predict astronomical positions and eclipses. In the late 18th century, a chess-playing machine called The Turk Automaton Chess Player was exhibited in Europe and America. The machine became very famous because it was able to beat many opponents, but it was exposed as a hoax in the early 1820s – a chess player was hidden within its cabinet.
Between 1940 and 1950, Isaac Asimov wrote a collection of short stories (I, Robot) about the interaction of humans, robots, and morality. His writings introduced Asimov's Three Laws of Robotics: 1) A robot may not injure a human being or, through inaction, allow a human being to come to harm. 2) A robot must obey the orders given it by human beings except where such orders would conflict with the First Law. 3) A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. Asimov's writings predated the availability of commercial computers.
Samuel's checker player
In 1959, after the first commercial computers became available, Arthur Lee Samuel, an early researcher in artificial intelligence, developed a computer program to play checkers. His program had a scoring function based on the position of the board at any given time and had various mechanisms by which the program could learn and become better, such as reevaluating the scoring function based on input from professional games and by playing thousands of games against itself. The program achieved sufficient skill to challenge good players.
in 1968, the science fiction film 2001: A Space Odyssey was released. It was based on a short story by Arthur C. Clarke. The film depicted space flight to Jupiter in a spaceship controlled by a computer named HAL which was endowed with artificial intelligence. During the journey, HAL, who was not programmed with Asimov's laws, decided that the mission was too important to entrust it to the human crew and the machine started killing them one by one. The film highlighted the difficulties that the human race may encounter with smart computers and the possibility that intelligent machines may eventually eliminate humans.
Research topics in Artificial Intelligence
A computer program has to be able to reason, plan and learn in order to achieve artificial intelligence. The interaction of the computer with the outside world and the limitations of the computer hardware also need to be considered. Research in artificial intelligence often has to deal with optimum methods of storing and retrieving information so that the computer can respond quickly. The following is a list of important topics in artificial intelligence.
Reasoning the the ability to draw logical conclusions from the available information. The basis of reasoning in AI is based in mathematics and inductive and deductive logic. Some examples of reasoning are:
Planning is the process of determining all the tasks needed to accomplish a goal. For example, to buy some sugar, you may need to get your car keys and your wallet, start the car, drive to the store, find and obtain the sugar, pay for the sugar and bring it back home. Planning takes into account rules that determine when certain tasks can or cannot happen, such as getting the car keys before being able to start the car.
Machine learning, like human learning, involves remembering the outcomes of previous experiences in order to select the most favorable results or to avoid problems in the future. Many techniques have been explored by researchers of machine learning, but they generally involve storing information about an action and its result and recalling the information at a later time to guide future actions.
Information storage and retrieval is a fundamental aspect of AI. Knowledge representation is the way in which information is configured in order to be able to store it, retrieve it and manipulate it. Some aspects of knowledge representation involve data compression, organization and indexing to speed up retrieval, and data formats for storing sounds, images and text.
Natural Language Processing (NLP)
Handling human languages has been one of the most sought-after goals of artificial intelligence. The nuances and ambiguities of language pose formidable problems for researchers. Idioms like "cut the mustard" and "beat around the bush" cannot always be translated into other languages without a lot of explanation.
Linguistic analysis generally relies on a lexicon (dictionary) where the words contain syntactic and semantic attributes that can be used for disambiguation. Morphological analysis is used to obtain the roots and suffixes of the words. Syntactic analysis or grammar determines the relationship between the words and identifies well-structured grammatical units. Semantic analysis puts together the syntactic units to deduce the meaning of the text. For text searching applications, the semantic units may be supplemented with synonyms so that a query for "bird flu" can retrieve documents about "avian influenza".
Natural language is the fundamental way in which humans will interface with AI devices. Advances in NLP will benefit AI and automatic translation, which is very important for global marketing by multinational organizations.
Manipulation and Control of Objects
A computer with artificial intelligence could take a spoken or written question and provide the answer on a screen. However, we expect AI computers to do more than that. We want our computers to turn on the coffee pot in the morning, regulate the temperature in our house and warm up the car when we are ready to go to work.
Robots are in fact mobile AI computers. We could say that robots include airplanes with autopilots that can control takeoff, navigation and landing. Self-driving cars such as those developed by Google can also be considered AI robots. Robot technology generally incorporates collision avoidance to prevent accidents.
Advances in AI
1966 - ELIZA was a chatterbot developed by Prof. Joseph Weizenbaum that used simple pattern matching techniques on input text and responded with questions that gave the impression that the computer understood the topic. This was an early example of Natural Language Processing.
1997 - Deep Blue was a chess-playing computer developed by IBM. It was the first AI program to win both a chess game and a chess match against Garry Kasparov, a reigning world champion under regular time controls.
2011 - Watson is a question answering computer system developed by IBM capable of answering questions posed in natural language. Watson competed on Jeopardy! against former winners Brad Rutter and Ken Jennings. Watson received the first place prize of one million dollars.
Personal Assistants - Companies are very actively developing computer interfaces that can answer natural language questions, execute user commands and serve as personal assistants. Siri for Apple devices was introduced in 2011. Google Now for Android devices performs user-initiated queries and proactively delivers information that users may want based on their previous search histories. Microsoft introduced Cortana as an intelligent personal assistant for the Windows Phone and Windows 10. The wide use of smart phones and devices connected to the Internet is accelerating the rate at which intelligent software is being developed for languages such as English, French, German, Italian, Spanish, Chinese and Japanese.
The Turing test determines a machine's ability to exhibit intelligent behavior equivalent or indistinguishable from that of a human. In 1950, Alan Turing proposed a test consisting of a human evaluator, a machine designed to generate human-like responses, and a human participant. All three would be separated from one another, and the evaluator would try to determine which of the other two was the machine by asking questions using a computer keyboard. If the evaluator cannot reliably tell the machine from the human, the machine would pass the test.
Consciousness and self-awareness
Many scientists believe that, when more powerful AI computers are developed, the machines will eventually become self-aware. Ray Kurzweil, a computer scientist and inventor, believes that computers will be able to pass the Turing test by 2029, and that by the year 2045 there will be a technological singularity where non-biological computation will exceed the capacity of all living biological human intelligence. This will result in a disruptive transformation of society. Critics have disputed many of Kurzweil's predictions, but at the rate at which AI technology is developing, it is very likely that computers with human-like capabilities will be developed during the 21st century.
Transhumanism is an international and intellectual movement that aims to transform the human condition by developing and creating widely available technologies to greatly enhance human intellectual, physical, and psychological capacities. Transhumanists believe that human beings may eventually be able to transform themselves into different beings with abilities so greatly expanded from the natural condition that they could be called posthuman beings. It is difficult to differentiate between such beings and cyborgs. A cyborg is an organism that has restored function or enhanced abilities due to the integration of some artificial components that lengthen lifespan and enhance biological functions. There are many examples in contemporary medicine that already fit this description, such as cochlear implants, artificial limbs, pacemakers, motorized wheel chairs, and many others. Just by using a smart phone, a person has access to the Internet and can answer many questions that would be impossible without the artificial assistance.
Machine independence from humans
Once machines have artificial intelligence and the machines have consciousness, there will be a competition for resources between the machines and humans. The smart machines will need a source of energy to continue to function. This will be an acceleration of the competition that already exists for foods such as corn and sugar that are used to produce fuel for automobiles.
AI machines will need to develop self-preservation, as in Asimov's third law, but the machines will not necessarily respect humans. Self-preservation will require avoiding dangers that would compromise operation of the machines. AI machines will need to learn to protect themselves from malicious cyberattacks or physical attacks, including attacks from other machines or humans who oppose the new technology.
Replication of intelligent machines will initially consist of cloning the software and memory banks of existing machines to other machines with the same hardware capabilities. The replication of robots is much more complex because it will require setting up factories and assembly plants for the hardware. Eventually, intelligent machines could run the factories, culminating in the technological singularity predicted by Kurzweil.
SpaceX founder Elon Musk, physicist Stephen Hawking, Apple co-founder Steve Wozniak and other prominent people in technology, robotics, and artificial intelligence have warned against the military use of AI weapons because this could start a global arms race to produce autonomous killing machines. Smart bombs with the appearance of ordinary household items, but fitted with speech recognition and face recognition could be programmed to detonate only when a particular individual is in close proximity. Targeted killings of terrorist leaders could be the first applications of such AI weapons.