1.1 Acting Like Humans

In1950,AlanTuringproposedan operationaldefinition(ratherthan alist of sufficient and necessary conditions) of intelligence, later named the Turing Test. An entity is deemed intelligent, according to Turing, if it demonstrates an ability to achieve human-level performance in all cognitive tasks, sufficienttofool aninterrogator. Turingpredicted thatby theyear2000, computers wouldbe able to pass the Turing test. ²

ELIZA, an early AI system intended as a therapist, fails the Turing Test.

PERSON: My wife says I’m often depressed. ELIZA: I’m sorry to hear you are depressed. PERSON: The truth is I am quite unhappy. ELIZA: Do you think coming here will help you not to be unhappy? PERSON: Well, perhaps by talking to you I will find out why I

hate my brother. ELIZA: Tell me more about your family.

Like the word brother in this example, a sentence such as I admire Mother Teresa triggers the response Tell me more about your family.

Designing a system that can simulatehumanbehavior(i.e.,pass theTuring test) is not theholy grail ofAI research,justassimulating the flightofbirdswasnottheoverarchinggoaldriving the work of the Wright brothers.

1.2 Thinking Like Humans

In1963,AllenNewell andHerbSimondesigned theGeneralProblemSolver(GPS), which wasintended tobe aprogram that simulatedhuman thought. The nameGPSderivedfrom theprogram’s architecture, whichdistinguishedbetweengeneralknowledgeabout reasoning andknowledgeabout specific problem domains. GPS used means-end analysis in its search for solutions.

QuotingAristotle,

² More specifically, he predicted that the average interrogator would not be able to distinguish a computer from a human more than 70 per cent of the time, after a five minute conversation.

We deliberate not about ends, but about means. For a doctor does not deliberate whether he shall heal, nor an orator whether he shall persuade . . . They assume the end and consider how and by what means it is attained, and if it seems easily and best produced thereby; while if it is achieved by one means only they consider how it will be achieved by this and by what means this will be achieved, till they come to the first cause, . . . and what is last in the order of analysis is first in the order of becoming.

In today’s parlance, means-end analysis assumes an initial state and a goal state, computes the difference between them, and if that difference is not zero, searches for ways to minimize it by replacing the goal with a sufficient set of subgoals, and then recurring.

By comparing GPS traces with those of human subjects, Newell and Simon discovered that the behavior of GPS was largely a subset of human behavior. Today, the study of human cognition characterizes the field of cognitive science, rather than AI.

1.3 Thinking Logically

The Laws of Thought approach toAI relies onpatternsfor argument structure rooted inAristotle’s syllogisms (e.g., All men are mortal; Socrates is a man; therefore, Socrates is mortal). In the late 1800’s and early 1900’s, the formal logic movement was advanced by Guiseppe Peano, George Boole, Gottlob Frege, Alfred Tarski, Kurt G¨odel, and others. Perhaps inspired by early progress, DavidHilbertbecame aproponentof a school of thoughtknown as logicism, or formalism. Thegoal of this program was to devise a logic, or formal system, capable of deriving all mathematical theorems, thereby uncovering all possible mathematical intuitions. Ultimately, G¨odel’s Incompleteness Theorem(1931), which statesthatinsufficientlypowerfullanguagesthereareunprovabletruths, served to dismantle the logicist/formalist program.

1.4 Acting Rationally

Modern AI can be characterized as the engineering of rational agents. An agent is an entity that

(i) perceives, (ii) reasons, and (iii) acts. In computational terms, that which is perceived is an input; to reason is to compute; to act is to output the result of computation. Typically, an agent is equipped with objectives. A rational agentisonethat actsoptimally with respecttoitsobjectives.

Agents are often distinguished from typical computational processes by their autonomy. They operate without direct human intervention. In addition, agents are reactive—they perceive their environments, and attempt to respond in a timely manner to possibly changing conditions—and proactive—their behavior is goal-directed, rather than simply response-driven.

Agent Sensors Actuators

Human	Senses	Arms, Legs, etc.
Robotic	Cameras	Motors, Wheels, etc.
Software	Bit Strings	Bit Strings

Table 1: Examples of Agents. Examples of agents include human agents, robotic agents, and software agents. (See Table 1.) 3

Perception

ENVIRONMENT

Actuation

Figure 1: Intelligent Agents = Perception + Reason + Actuation.

Autonomous agents may be rule-based, goal-based, or utility-based. Rule-based agents operate according to hard-coded sets of rules, like ELIZA. A goal-based agent acts so as to achieve its goals, by planning a path from its current state to a goal state, like GPS or theorem provers. Utility-based agents distinguish between multiple goals based on their respective utilities.

Some dimensions of agent environments include:

• Deterministic vs. Nondeterministic: is the next state predictable (e.g., chess), or is there uncertainty about statetransitions(e.g., backgammon)?
• Discretevs.Continuous: cantheenvironmentbedescribedindiscreteterms(e.g., chess), or isthe environment continuous(e.g., driving)?
• Static vs.Dynamic: is the environment static(e.g., chess), or can it change while the agent is reasoning aboutitsplan of action(e.g., driving)?
• Sequential vs. One-shot: does the agent need to reason about the future impact of its immediate actions(e.g., chess), or canittreat each actionindependently(e.g., Rochambeau)?
• Single agent vs. Multiagent: can we assume the agent operating alone in its environment, or need it explicitly reason about the actions of other agents (e.g., chess, backgammon, Rochambeau, driving)?