The document provides an introduction to artificial intelligence (AI). It discusses the goals of understanding intelligent behavior and building intelligent agents. It then examines four perspectives on defining AI: (1) acting humanly vs thinking rationally, and (2) focusing on thought processes vs behavior. Each perspective is associated with different approaches to AI like the Turing Test, cognitive modeling, laws of thought, and rational agents. The document also outlines some current capabilities of AI like robotic vehicles, speech recognition, game playing, and machine translation. It introduces the concepts of agents, percepts, and rational agents that aim to maximize performance. Finally, it categorizes different types of environments that agents can operate in.
2. AI >> Goals
Ambitious goals >>
- understand âintelligentâ behavior
- build âintelligentâ agents
What is Intelligence?
- capacity to learn and solve problems
- the ability to act rationally
2
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
3. What is AI?
Definitions of AI fall into 4 different perspectives.
â¸Two dimensions:
i) Thinking vs Acting
ii) Human vs Rational
â¸The top row is concerned with the thought process and reasoning
â¸The bottom row addresses behavior.
â¸Human centered approach is a part of empirical science, involving observations and hypotheses about human behavior,
measuring success in terms of fidelity to human behavior
â¸The rationalist approach involves a combination of mathematics and engineering.
3
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
2. Thinking Humanly 3. Thinking Rationally
1. Acting Humanly 4. Acting Rationally
Thought
Behavior
Human-like Intelligence Pure Rationality
4. 1. Acting Humanly >> Turing Test approach
The art of creating machines that perform functions that require intelligence when performed by people.
Turing Test >>
â¸Proposed by Alan Turing (1950), was designed to provide a satisfactory operational definition of intelligence.
â¸A computer passes the test if a human interrogator, after posing some written questions, cannot tell whether the written
responses come from a person or from a computer.
4
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
5. 1. Acting Humanly >> Turing Test approach
â¸To design such computers to pass the Turing Test need to possess the following capabilities:
âšNatural Language Processing â to enable it to communicate successfully in English.
âšKnowledge Representation â to store what it knows or hears.
âšAutomated Reasoning â to use the stored information to answer questions and to draw new conclusions.
âšMachine Learning â to adapt to new circumstances and to detect and extrapolate patterns.
â¸To pass the Total Turing Test computers will also need:
âšComputer vision â to perceive objects.
âšRobotics â to manipulate objects and move about.
â¸AI researcher have devoted little effort to passing the Turing Test, believing that it is more important to study the underlying
principles of intelligence than to duplicate an exemplar.
5
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
6. 2. Thinking Humanly >> Cognitive Modeling approach
The automation of activities that we associate with human thinking, activities such as decision-making, problem solving,
learning etc.
If we want to say that a given program thinks like a human, then we must have some way of determining how human think.
There are 3 approaches:
â¸through Introspection â trying to catch our own thoughts as they go by.
â¸through Psychological experiments â observing a person in action.
â¸through Brain imaging â observing the brain in action.
Once we have sufficiently precise theory of the mind it becomes possible to express the theory as a computer program.
>> If the programâs input-output behavior matches corresponding human behavior, that is evidence that some of the
program's mechanisms could also be operating in humans.
6
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
7. 2. Thinking Humanly >> Cognitive Modeling approach
The interdisciplinary field of cognitive science brings together computer models from AI and experimental techniques from
psychology to construct precise and testable theories of the human mind.
Cognitive Modeling Approach >>
â¸General Problem Solver(GPS) developed by Allen Newell and Herbert Simon in 1961 is a computer program that is
designed to solve problem correctly.
â¸The scientists were more concerned with comparing the trace of its reasoning steps to traces of human subjects solving
the same problems.
Real cognitive science, however is necessarily based on experimental investigation of actual humans or animals. So we will
not consider this field assuming that we have only computers available for experimentation.
7
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
8. 3. Thinking Rationally >> Laws of Thought approach
The study of the computations that make it possible to perceive, reason, and act.
â¸Aristotle: one of the first to attempt to codify âright thinkingâ.
For example,
âSocrates is a man; all men are mortal; therefore, Socrates is mortalâ
â¸These laws of thought approach were supposed to govern the operation of the mind; their study initiated the field called
logic.
â¸Logicians in the 19th century developed various forms of logic: notation and rules of derivation for thoughts;
â¸Problems:
âšIt is not easy to take informal knowledge and state it in the formal terms required by logical notation, particularly
when the knowledge is less than 100% certain.
âšThere is a big difference between solving a problem âin principleâ and solving it in practice.
8
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
9. 4. Acting Rationally >> Rational Agent appraoch
Agent >> An agent is just something that acts. All computer programs do something but computer agents are expected to do
more: operate autonomously, perceive their environment, persist over a prolonged time period, adapt to change, and create
and pursue goals.
Rational Agent >> It is one that acts so as to achieve the best outcome or, when there is uncertainty, the best expected
outcome.
Advantages over other approaches:
âšIt is more general than the âlaws of thoughtâ approach because correct inference is just one of several possible
mechanisms for achieving rationality.
âšIt is more amenable to scientific development than are approaches based on human behavior or human thought
because the standard of rationality is mathematically well defined and completely general, and can be unpacked to
generate agent designs that provably achieve it.
9
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
11. State of the Art >> What can AI do today?
â¸Robotic vehicles
âšSTANLY : A driverless robotic car sped through the rough terrain of the Mojave dessert at 22mph, finishing the 132
mile course first to win the 2005 DARPA Grand Challenge.
âšCMUâs BOSS : It won the Urban challenge, safely driving in traffic through the streets of a closed air force base,
obeying traffic rules and avoiding pedestrians and other vehicles.
â¸Speech recognition
âšcalling United Airlineâs to book flight can have the entire conversation guided by an automated speech recognition
and dialog management system.
â¸Autonomous planning and scheduling
âšA hundred million miles from Earth, NASAâs Remote Agent program became the first on-board autonomous
planning program to control the scheduling of operations for a spacecraft.
â¸Game playing
âšIBMâs DEEP BLUE became the first computer program to defeat the world champion in a chess match when it
bested Garry Kasparov by a score of 3.5 to 2.5 in an exhibition match.
11
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
12. State of the Art >> What can AI do today?
â¸Spam fighting
âšEach day learning algorithms classify over a billion messages as spam, saving the recipient from having to waste
time for deleting.
â¸Logistic planning
âšIn 1991, US forces deployed a Dynamic Analysis and Re-planning Tool to do automated logistic planning and
scheduling for transportation.
â¸Robotics
âšThe iRobot Corporation has sold over two million Roomba robotic vacuum cleaners for home use.
â¸Machine translation
âšA computer program automatically translates from Arabic to English
These are just a few examples of artificial intelligence systems ⌠⌠âŚ
12
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
13. Agents
An agent is anything that can be viewed as
- perceiving its environment through sensors and
- acting upon that environment through actuators
Human Agent >>
eyes, ears, and other organs for sensors;
hands, legs, mouth, and other body parts for actuators.
Robotic Agent >>
cameras and infrared range finders for sensors;
various motors for actuators.
Software Agent >>
keystrokes, file contents, and network packets as sensory inputs;
display the screen, writing files, and sending network packets as actions.
13
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
14. Agents
Percept >> agentâs perceptual inputs at any given instant.
Percept sequence >> complete history of everything the agent has ever perceived.
Agent function >> maps any given percept sequence to an action.
[f: P* ď A]
Agent program >> runs on the physical architecture to produce f.
Agent = Architecture + Program
14
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
15. Vacuum-cleaner World
15
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Percepts >> location, status e.g. [A, Dirty]
Actions >> Left, Right, Suck, NoOp
16. Vacuum-cleaner World >> Tabular Agent Fn
16
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Right way to fill out the table??
What makes an agent good or bad,
intelligent or stupid?
17. Rational Agents
Performance Measure >> An objective criterion for success of an agentâs behavior.
E.g. performance measure of a vacuum-cleaner agent could be amount of dirt cleaned up, amount of
time taken, amount of electricity consumed, amount of noise generated, etc.
Rational Agent >> For each possible percept sequence, a rational agent should select an action that is
expected to maximize its performance measure, given the evidence provided by the percept sequence
and whatever built-in knowledge the agent has.
17
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
18. Task Environment >> PEAS
In designing a rational agent, the first step must always be to specify the task environment as fully as
possible.
PEAS >> Performance measure, Environment, Actuators, Sensors
Example: agent for a self-driving car
18
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
19. Task Environment >> PEAS
19
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
20. Environment Types
Fully/Partially Observable >>
Is everything an agent requires to choose its actions available to it via its sensors?
- If so, the environment is fully accessible = Fully Observable
- If not, parts of the environment are inaccessible = Partially Observable
- Agent must make informed guesses about world
- If the agent has no sensors at all = Unobservable
20
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Fully Observable Partially Observable
⢠Crossword puzzle
⢠Chess with a clock
⢠Backgammon
⢠Poker
⢠Taxi driving
⢠Part-picking robot
21. Environment Types
Single/Multi-agent >>
An agent operating by itself in an environment or there are many agents working together.
21
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Single Agent Multi-agent
⢠Crossword puzzle
⢠Part-picking robot
⢠Chess with a clock
⢠Poker
⢠Backgammon
⢠Taxi driving
22. Environment Types
Deterministic/Stochastic >>
If the next state of the environment is completely determined by the current state and the action executed
by the agent, then we say the environment is deterministic; otherwise it is stochastic.
22
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Deterministic Stochastic
⢠Crossword puzzle
⢠Chess with a clock
⢠Poker
⢠Backgammon
⢠Taxi driving
⢠Part-picking robot
23. Environment Types
Episodic/Sequential >>
Is the choice of current action dependent on previous actions?
- If not, then the environment is episodic
In non-episodic/sequential environments:
- Agent has to plan ahead cause current choice will affect future actions
23
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Episodic Sequential
⢠Part-picking robot ⢠Crossword puzzle
⢠Chess with a clock
⢠Poker
⢠Backgammon
⢠Taxi driving
24. Environment Types
Static/Dynamic/Semi-dynamic >>
- Static environments donât change while the agent is deliberating over what to do.
- Dynamic environments do change while the agent is deciding on an action. So agent should/could
consult the world when choosing actions. Alternatively: anticipate the change during deliberation OR make
decision very fast.
- Semi-dynamic: If the environment itself does not change with the passage of time but the agent's
performance score does
24
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Static Semi-dynamic Dynamic
⢠Crossword puzzle
⢠Poker
⢠Backgammon
⢠Chess with a clock ⢠Taxi driving
⢠Part-picking robot
25. Environment Types
Discrete/Continuous >>
If the number of distinct percepts and actions is limited, the environment is discrete, otherwise it is
continuous.
25
Mohammad Imam Hossain | Lecturer, Dept. of CSE | UIU
Discrete Continuous
⢠Crossword puzzle
⢠Chess with a clock
⢠Poker
⢠Backgammon
⢠Taxi driving
⢠Part-picking robot