Artificial Intelligence
From Isopedia
Artificial Intelligence (AI) is a branch of computer science that is concerned with the development of intelligent machines that are able to continuously adapt and learn.
Contents |
History of Artificial Intelligence
1950's
Interest in artificial intelligence sparked in the late 1940’s. Many believe Alan Turing, an English mathematician, was the first intellect to address artificial intelligence (AI). In 1950, Turing developed the Turing test, a test that determines if a machine should be considered intelligent. A human converses with both a human and a machine, via text, and has to choose which is which. If the human cannot correctly tell, the machine is considered intelligent. The first conference on artificial intelligence was in 1956 at Dartmouth College, New Hampshire, where the term “artificial intelligence” was conceived. This is where the first running AI program, the Logic Theorist (LT), was demonstrated by Allen Newell, J.C. Shaw and Herbert Simon. The Dartmouth Conference further ignited interest in AI. Centers for AI research were soon founded at Carnegie Mellon, MIT, IBM and Stanford. The first well-known AI programs were developed to play board games, such as chess and checkers. In 1958, John McCarthy designed LISP, a programming language for artificial intelligence. LISP, which stands for list programming, is still used today.
1960's-1970's
In 1963, the U.S. Department of Defense’s Advanced Research Projects Agency granted MIT $2.2 million to research AI. This drew more computer programmers to the US. Continued funding and more computer programmers helped AI develop rapidly in the 1960’s. The AI Lab at Stanford University designed Shakey, a robot that could follow easy instructions and move around blocks. Much attention was drawn to AI when Stanley Kubrick made the film 2001: A Space Odyssey, in which a computer murders many people on board a ship. In the 1970’s, the popularity of AI dropped. Irrational expectations for AI were not met, and funding was dropped. AI continued to be developed, however, and actual applications of AI in the real world were founded. One of these is ‘fuzzy control’, which used fuzzy logic to process control. This was applied to steam engines in the ‘70s.
1980's-Today
Fortunately, in the 1980’s, more applications of AI were found, and AI hardware and software became cheaper and easier to make. LISP machines were developed, and marketed for the general public. Expert systems, systems that make expertise information available to decision makers and technicians quickly, were developed. Over $400 million of revenue was made off of these AI products. Speech recognition programs and expert systems were developed in the 1980’s. Artificial intelligence was being used more and more in the military. The costs spent by the government on AI in the 1950’s were completely offset by the costs saved by using the AI methods in the Persian Gulf War. In the 1990’s, AI programs were sought out to help solve problems quickly and more efficiently. AI software was integrated with already existing programs, and used in many industries. Credit card fraud detection systems, face recognition systems, and business revenue and staffing requirements predictions systems are a few examples of AI integrated programs. Web crawlers, Web-based programs that use AI to extract information, became essential on the Internet. In the 2000’s, many toys, such as ‘smart pets’, were developed using AI. Today, more applications are being sought for AI, as AI continues evolving and finding more uses.
Conventional Artificial Intelligence
Conventional Artificial Intelligence is the first of the two branches of Artificial Intelligence and is also known as Symbolic AI, Logical AI, Neat AI, and Good Old Fashioned AI. Methods of Conventional AI include expert systems, case based reasoning, Bayesian networks, and behavior based AI. These methods are now classified as machine learning. Machine learning concerns the question “How can we program systems to automatically learn and to improve with experience?”
Machines learn whenever they change structure, programs, or data, which then allows for improvement in their future performance. Computers are given a task to do, the task is repeatedly demonstrated in the same manner as how the task should be performed, and the computer learns by example. It is an outgrowth of a mix of Computer Science and Statistics, and many algorithms have been found to have direct relationships with statistics. The method involves the development of algorithms and techniques that help computers to learn and attempts to automate parts of the scientific method. Some systems try to eliminate human interaction as much as possible, while others use a mix of both human and machine interaction. With the use of machine learning, computers can produce programs that would have never been able to be written by people.
Machine learning is beginning to become a vital part in the three areas of the software world: Data mining, difficult-to-program applications, and customized software applications.
1. Data mining: The process of using historical databases to improve subsequent decision making.
2. Difficult-to-program applications: Machine learning algorithms take care of applications that have been found too difficult to program by hand.
3. Customized software applications: Enables software to automatically cuztomize itself to individual users.
Researchers say that as technology grows, the need for machine learning grows. Better algorithms and faster computers will allow endless possibilties for machine learning uses. People are looking for software to adapt itself, as software becomes larger and less manageable.
Applications of Conventional AI
Conventional AI is now used in such activities as natural language processing, search engines, medical diagnosis, detecting credit card fraud, stock market analysis, speech and handwriting recognition, game playing, and robot locomotion. New algorithms have even learned how to steer cars autonomously driving at 70 miles an hour an public highways and have learned the reading interests of individuals to create personally customizes electronic newspapers.
Computational Intelligence
Computational Intelligence (CI) is one of the two main branches of Artificial Intelligence (AI). The main purpose of CI is to understand the principles that make intelligent behavior possible- This includes in both natural and artificial systems.
Computational intelligence research either rejects statistical methods (as is the case with fuzzy systems), or ignores statistics (as is the case with most neural network research). In contrast, machine learning research rejects non-statistical approaches to learning, adaptivity, and optimization.
- Fuzzy Systems
- Fuzzy logic is a superset of conventional logic that has been extended to handle the concept of "partial truth" (truth values between "completely true" and "completely false"). A fuzzy system is one that uses a collection of fuzzy logic rules to reason about data.
- Neural Network Research
- A computer mechanism that is closely modeled in the structure of the human brain. It's models can be used to either simulate or analyze complex occurances.
CI is centered around the belief that reasoning is computation (hence the name Computational Intelligence).It is the study of the design of intelligent agents. An agent is anything that acts within and environment. This includes things such as individual humans, and ranges up to human society, which is probably the most intelligent known agent. An agent could be something from a robot, to an infobot.
At any given time an agent has the following characteristics
1. prior knowledge about the world
2. past experience that it can learn from
3. goals that it must try to achieve or values about what is important
4. observations about the current environment and itself
5. using these imputs, it completes some task
There is some tension in CI between the science portion, trying to understand the principles behind reasoning, and the engineering portion, building programs to solve specific problems. CI theories are about how interesting problems can be represented and solved through machinary. It could also be referred to as "computational epistemology" because it can be seen as a method of studying the problem of the nature of knowledge, but with a more powerful tool than was possible prior to the use of computational intelligence.
CI is closely related to the field of computer science. Most CI research is done within this department. It is essential to understand algorithms, data structures, and combinational complexity in order to build intelligent machines.
Applications of Computational Intelligence
CI has many diverse applications: ranging from medical diagnosis, scheduling factory processes, robots for hazardous environments, chess playing, autonomous vehicles, natural language translation systems, and cooperative systems.
There are three main application domains. The first is the autonomous delivery robot, that complete tasks such as delivering packages and coffee to people in the building. The second is the diagnostic assistant. This helps a human troubleshoot problems and suggests repairs to solve the problem. The "infobot" is something that can search for information on a computer system.
Artificial Intelligence in the Field of Business
Artificial Intelligence is currently used in many different types of businesses all over the world. Examples of AI are found in banking and investing firms, manufacturing plants, law enforcement agencies, hospitals, as well as data mining and other internet uses to name a few. One major advantage and reason for the use of Artificial Intelligence in the business world, specifically the work place is to accomplish the jobs and tasks that require repetitive actions or might be thought of as a degrading (Wikipedia 2006).
Other specific examples of Businesses using Artificial Intelligence
A) General Motors roughly uses 16,000 robots to complete various tasks on the automotive production line.
B) Cell Phone companies use A.I. technology for voice recognition functions.
C) Banks use A.I. programs to monitor various types of transactions for fraud.
D) Some lifeguards use “vision-recognition” programs to scan swimming areas for drowning victims.
E) Hewlett Packard has developed an A.I. Dj that can sense the reactions of the audience and adjust the playlist accordingly.
Current A.I. technology attempts to “connect people to computers, people to knowledge, people to the physical world, and people to people” (http://people.csail.mit.edu/phw/briefing.html). This ideology is possible due to the latest technological advances and largely through the inter-connectedness of the internet. Businesses specifically have been seeking more A.I. options as a means to gain a competitive advantage; lower operating costs, as well as enter new areas for greater revenue. An incredible advantage of A.I. technology is that it is able to span so many different areas in the business world as well as “the real world”.
Artificial Intelligence and Ethical Issues
Ethics is a major concern to the development of artificial intelligence. Although there are many benefits for the use of AI, it also brings many costs along with it. Some of these costs provoke ethical concerns. These concerns include employment, medicine, morality, and goverment.
One of the strongest fears that AI evokes from people is the fear that AI will take jobs from humans, leaving many unemployed. Computers already have the ability to complete certain tasks that were formally done by humans, such as account keeping, inventory, production in factories, and even entertainment. We can already see the effect that machines have had on the car manufacturing industry. The automation of assembly lines have left many factory workers unemployed. Many more industries and businesses will be automated in the future, leaving even more people without a job.
Another ethics issue is in the medical field. Although computers can sometimes more accurately diagnose and cure patients, even computers can make mistakes. If artificial intelligence misdiagnoses a patient, who is at fault? The doctor, the programmer, or the machine?
Fear of AI taking control is one of the most popular ethical concerns today that society has grasped. The media has fostered this fear in movies such as The Terminator, The Matrix and I, Robot. People fear that with the development of AI, machines will be able to surpass human abilities, and therefore become the dominant “race.” In the popular Matrix Trilogy, the human race is imprisoned by machines and used as batteries. Computers have already surpassed the human ability in chess, since IBM designed a successful computer program called Deep Blue in 1997.
Can machines have emotions and a conscience? This is another question often asked in the AI field. Critics and supporters alike have views on this subject. Some believe a machine is as imperfect as its maker. Others believe machines can learn from humans and possibly even surpass them morally. Some AI supporters hope to see AI used in politics, to make un-biased decisions.
The possible good that AI can achieve should not be overlooked. Artificial intelligence has the possibility to save lives through medicine, war, increasing safety, and replacing humans in dangerous jobs. Intelligent machines help in engineering safe and efficient buildings for people. The real dilemma is deciding whether the benefits outweigh the costs.
Sources
http://www.aaai.org
http://www.wikipedia.org
http://www-formal.stanford.edu/jmc/whatisai/node1.html
http://library.thinkquest.org/2705/history.html
http://www.charlesriver.com/resrcs/chapters/1584502789_1stChap.pdf
http://groups.csail.mit.edu/medg/ftp/psz/SchwartzAnnals.html
http://www.narin.com/attila/ai.html
http://www.cs.ubc.ca/spider/poole/ci/ch1.pdf
http://www.webster.com/dictionary/artificial%20intelligence
http://www.cs.cmu.edu/Groups/AI/html/faqs/ai/fuzzy/part1/faq.html
Team Members
Kristin Dupre
Dan Gorman
William Sonnemann
Jocelyn Touchette
Katie Wilusz
