1. Brief history

Computer appeared in the 1940s. Though initially it was used for numerical calculation, soon people realized that it can carry out many other intellectual activities by manipulating various types of symbols. Naturally, people began to wonder whether all mental activities can be carried out by computer, and if not, where is the boundary.

In 1950, British mathematician and computer scientist Alan Turing published an important article "Computing machinery and intelligence". Its major contents are:

• An "Imitation Game" (later called "Turing test") was proposed as the standard of intelligence, or thinking, in which the intellectual capacity of a system is separated from it physical capacities and appearance.
• Various types of objections to the possibility of a thinking machine were analyzed and rejected.
• The possibility of a "thinking" computer was discussed and it was suggested that it can be achieved by building a learning system.

In the early days, people were generally optimistic about AI. For example, a "General Problem Solver" was programmed with the hope that it can solve all kinds of problems (when properly represented).

Soon, researchers found various kinds of issues which made them to turn to domain-specific knowledge for help. Consequently, in the late 1970s and early 1980s the first generation of "Expert System" appeared, with commercial success in limited domains.

With the problem of the traditional "Symbolic AI" approach became more and more clear, various types of alternative approaches became popular in the late 1980s. As a result, today's AI is a domain where many research paradigms co-exist and compete with each other.

Reference: a brief history of Artificial Intelligence, Dartmouth Artificial Intelligence Conference: The Next Fifty Years (AI@50).

2. Fields in AI

By the scope of cognitive/intelligent capacities under research, the fields include:

• core cognitive facilities, such as searching, reasoning, learning, planning, categorizing, and recognizing;
• input/output facilities, such as sensorimotor managing, and natural language processing.

• rule-based system,
• case-based system,
• neural networks,
• genetic programming,
• logic programming (e.g., Prolog),
• functional programming (e.g., Lisp),
• ... ...

• game playing,
• theorem proving,
• data mining,
• ... ...

3. Nature of the domain

The difficulty comes not only from the simple fact that mind is one of the most complicated phenomena in the universe, but also from the nature of AI, which must be, at the same time,

• a branch of science, about how intelligence/mind/thinking work (so it is part of "Cognitive Sciences", with psychology, philosophy, logic, linguistics, neuroscience, and so on);
• a branch of engineering, about how to make new computer hardware and software.

1. a theory on intelligence (or part of it),
2. a formal model of the theory,
3. a computational implementation of the model.

4. AI and your future

AI for the minority of the students who really want to take it as career:

• Application opportunity (with a MS or BS degree): you can find jobs in expert systems, Prolog/Lisp programming, data mining, neural network, and so on, but the market is small and the jobs are not secure (though they can be interesting).
• Research opportunity (with a PhD or MS degree): interesting and challenging topics in many places, but you need to be well prepared for hard problems and tough career paths.

• get new ideas from the domain,
• follow the progress and be prepared for its future development,
• have a better idea about human thinking,
• have fun in thinking about the problems!