ECS 170:
Introduction to Artificial Intelligence
Lecture 1.
Prof. Rao Vemuri
Historical Background
Slide 1. Three Perspectives.
Artificial Intelligence can be approached from three intersecting viewpoints
Psychology
Engineering
Philosophy
Slide 2. Topics studied under AI
Artificial
Intelligence is the study of and simulation of intelligent behavior using
computers. Topics that fall under this broad category include:
Robotics: Artificial Intelligence is also
important in fields like robotics and cybernetics.
Slide 3. Activities Requiring Intelligence
Slide 4. Fields that can
contribute to AI
Slide 5. Characteristics of a typical AI Problem
Slide 6. Historical Evolution of AI
Slide 7. The Physical Symbol System Hypothesis
Newell and Simon's Physical Symbol System Hypothesis states:
"A physical symbol system has the necessary and sufficient means for intelligent action."
A physical symbol system "consists of a set of entities, called symbols, which are physical patterns that can occur as components of another type of entity called an expression (or symbol structure). Thus, a symbol structure is composed of a number of instances (or tokens) of symbols related in some physical way (such as one token being next to another). At any instant of time the system will contain a collection of these symbol structures. Besides these structures, the system also contains a collection of processes that operate on expressions to produce other expressions: processes of creation, modification, reproduction and destruction. A physical symbol system is a machine that produces through time an evolving collection of symbol structures. Such a system exists in a world of objects wider than just these symbolic expressions themselves."
"Two notions are central to this structure of expressions, symbols, and objects: designation and interpretation."
"Designation. An expression designates an object if, given the expression, the system can either affect the object itself or behave in ways dependent on the object. ... In either case, access to the object via the expression has been obtained, which is the essence of designation."
"Interpretation. The system can interpret an expression if the expression designates a process and if, given the expression, the system can carry out the process. ...Interpretation implies a special form of dependent action: given an expression the system can perform the indicated process, which is to say, it can evoke and execute its own processes from expressions that designate them."
Slide 8: Intelligent Agents (Use about 10 slides from Russel and Norvig with pictures here)
In a way, AI is the science of building intelligent agents
An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment with effectors.
Rational agent is one that does the "right thing". That is, whatever action it takes maximizes a performance measure. It does this by mapping percepts into actions.
An agent is autonomous to the extent that its behavior is determined by its own experience.
We will talk about four types of agents
Simple Reflex agents
Agents that keep track of the World
Goal-based agents and Utility based agents
Slide 20. Strong AI and Weak AI
The current general position is that in order for a system to be intelligent it must be capable of learning or "programming itself" and at the same time it probably must, in many cases, approach its world with just the right biases about how to learn or "program itself".
Weak Methods
AI attempted to respond to this problem of how to achieve intelligence without "programming" by "augmenting" the basic computational model defined by a Turing Machine. They did this by introducing the idea of a Weak Method. What is desired is to identify a perfectly general "program" or "method" that is as dumb as imaginable, but can provide the basis for "becoming more intelligent" through some learning mechanism.
The basic idea is referred to as generate and test. What this involves is being able to generate candidate
solutions to some problem and then being able to test whether a candidate solution is in fact the solution to the problem. For example, say you are given the problem 2x + 2 = 10 and asked to provide the value for x. Now, assume you know absolutely nothing about algebra; that is, you have no special "program" or method for solving this type of problem. What could you do?
Well, let's assume that you do know how to evaluate an arithmetic expression. Consequently, you can compute the value of an expression like 2(100) + 2 = 202. Given this ability to evaluate arithmetical expressions, you can then use that ability to specify a test. That is, you can see if 202 equals 10. Since it does not, you can conclude that 100 is not the solution to this problem. Now, you also need the ability to generate these candidate solutions.
In this case this ability to generate candidate solutions can be achieved by simply being able to generate a number, any number. So you need some sort of rule that generates a number, then you test that number to see if it is a solution. If the test succeeds, then stop and write out the solution. If not, simply generate another number, and so on. This is what is known as blind search. It is blind in the sense that there is not only no knowledge that is being used to direct the search, but also in the sense that the search is unsystematic. It is about the dumbest thing that we can imagine. The idea goes back to what is termed the British Museum Algorithm which gets its name from the notion that if you put a chimp or whatever in front of a typewriter, then eventually this chimp will generate all the books in the British Museum!
This brings us to the topic of search, which
we will take up soon.
Slide 21. Future Directions of AI
What is the direction in which future AI will evolve?
To be included in the "futures"
list, a research problem or application must have the following
characteristics:
1. Collaborative Community Effort: It must span several subfields of AI, requiring some degree of collaboration between AI researchers of different specialties. The idea is to help unify the fragmented subfields with a common purpose or purposes.
2. High Impact: It must address important problems of widespread interest. Solving the problem must matter to many people and not simply be adding another grain of sand on the anthill. This will help motivate and excite researchers, and justify the field to outsiders.
3. Short Horizon for Progress: It must be possible to have incremental progress and not be an all or nothing problem. For example, problems where we can reasonably expect to make significant measurable progress over the next 10 years or so.
4. Drive Basic Research: It should involve more than just applying current technology, but should drive basic research and the development of new technology (possibly in completely new directions).
In short, these problems should be "Grand
Challenges" for AI. If you were trying to describe the field of AI to a
layman, what concrete
problems would you use to illustrate the overall
vision of the field?
Saying that the goal of AI is to produce "thinking machines that solve problems" doesn't quite cut it.
Slide 22. Here is one list of "Grand
Challenges" in AI.
Department of Computer Science
University of California at Davis
Davis, CA 95616-8562