Chapter 1: Introduction

● Introduction to AI
○ History
○ Foundations

Contents ●
○ Applications and sub domains
Intelligent Agents
○ Agents, Environments, and their
types
○ Behavior and Performance
Measures
Introduction to AI
Intelligence
Intelligence involves certain mental activities composed of:

● Learning: ability to obtain and process new information

● Reasoning: being able to manipulate information in various ways
● Understanding: considering the results of information manipulation
● Grasping truths
● Seeing relationships and considering meanings …

Intelligence is the ability to apply knowledge to perform better in an environment

Artificial Intelligence (AI)
AI aims to create systems that can exhibit behavior and decision-making abilities
similar to those of humans.

Compare to human Compare to ideal concept of

performance intelligence

Thought process and reasoning Systems that think like humans Systems that think rationally

Behavior Systems that act like humans System that act rationally
Defining AI
Four ways to define AI:

1. Acting humanly
2. Thinking humanly
3. Thinking rationally
4. Acting rationally
Acting humanly: The Turing Test Approach
The Turing test was proposed by Alan Turing.

A computer passes the Turing test if a human interrogator, after posing some
written questions, cannot tell if the written responses come from a computer or from
a person.

The total Turing test requires interaction with objects and people in the real world.
Acting humanly: The Turing Test Approach
To pass the Turing test, the computer would need to
● communicate in a human language (Natural Language Processing)
● store what it knows or hears (Knowledge Representation)
● answer questions and draw conclusions (Automated Reasoning)
● adapt to new circumstances and to detect/extrapolate patterns (Machine
Learning)
To pass the total Turing test, a robot will need to
● perceive the world (Computer Vision and Speech Recognition)
● to manipulate objects and move around (Robotics)
Thinking humanly: The Cognitive Model Approach
● To say a program thinks like a human, we must know how humans think.
● Cognitive modeling approach relies on three techniques for learning human
thought
○ Introspection
Monitoring one’s own thought processes
○ Psychological experiments
Observing a person’s behavior given a set of circumstances, goals, resources, environments,
conditions etc.
○ Brain imaging
Observing the brain activity directly through mechanical means, e.g. CAT, PET, MRI etc.

Is it possible to make the machines think like human brain?

Thinking rationally: The “Laws of Thought” Approach
● A system is rational if it does the “right thing” given what it knows.
● The “Laws of Thought” approach models thinking as a logical process, where
conclusions are drawn (inference) based on some type of symbolic logic.
● Emphasis is on correct inference (i.e., thinking and reasoning in an ideal
world).
● Logic requires knowledge of the world that is uncertain. Theory of probability
allows modeling of uncertain world.
● Is rational thought, by itself, enough to generate intelligent behavior?
○ No. We need a theory of rational action.
Acting Rationally: The Rational Agent Approach
Agent = something that acts.

If we view ourselves as an agent, we realize that we have to be able to

● perceive the world (computer vision)

● perform actions in it (robotics)
● communicate with other humans/agents (language)
● have knowledge about the world, and draw conclusions/inferences and make
decisions based on the knowledge we have
● learn and adapt over time.
Acting Rationally: The Rational Agent Approach
Computer agents are expected to

● Operate autonomously
● Perceive their environment
● Persist over a prolonged period of time
● Adapt to changes
● Create and pursue goals

A rational agent is the one that acts so as to achieve the best outcome.
Acting Rationally: The Rational Agent Approach
All the skills required to pass the Turing test also allow an agent to act rationally.

AI has focused on the study and construction of agents that do the “right thing”.

The “right thing” is defined by the objective that we provide to the agent.
Artificial Intelligence
“The basis for what you expect from AI is a combination of how you define AI, the
technology you have for implementing AI, and the goals you have for AI.
Consequently, everyone sees AI differently.” - Massaron

“AI is the study of how to make computers do things which, at the moment, people
do better”. - Rich, Knight, Nair

“AI is Artificial Intelligence till it is achieved; after which the acronym reduces to
Already Implemented”. - Rich, Knight, Nair
History of AI
1943 Warren McCulloch and Walter Pitts introduced Boolean circuit model of brain (neural
network)

1950 Marvin Minsky and Dean Edmonds built the first neural network computer, called SNARC,
using vacuum tubes

1950 Alan Turing introduced the Turing test, machine learning, genetic algorithms and
reinforcement learning in his article “Computing Machinery and Intelligence”

1956 John McCarthy organized a 2-month summer workshop at Dartmouth College in’ Hanover,
New Hampshire to build a system that could do it all. No breakthroughs.
Allen Newell and Herbert Simon invented a mathematical theorem-proving system called
Logic Theorist (LT), and designed General Problem Solver (GPS)
History of AI
1952 - 1969: Early enthusiasm, great expectations
Early successes: LT, GPS, Arthur Samuel’s programs learned to play checkers at strong amateur level; his
program learned to play a better game than its creator.

1958 At MIT lab, John McCarthy defined the high-level language Lisp, which embodied the
central principles of knowledge representation and reasoning

Marvin Minsky moved to MIT; solved microworld problems, e.g., closed-form calculus
integration, geometry analogy problem etc.

1963 McCarthy started AI lab at Stanford. J. A. Robinson discovered a complete theorem-proving

algorithm for first-order logic

Overwhelming optimism: Predicting that AI would be solved in a matter of years

History of AI
1966 - 1973: Underwhelming results, AI winter
1957 Herbert Simon predicted that within 10 years, a computer would be a chess champion, and a
significant mathematical theorem would be proved by machine.

1966 Automatic Language Processing Advisory Committee (ALPAC) report led to the US
government to reduce its funding of the topic dramatically. This marked the beginning of the
first AI winter

1973 The Lighthill report led the British government to end support for AI research in all but two
universities.

1969 Minsky and Papert’s book “Perceptron” proved that they could not solve simple problems,
such as XOR. This book is largely credited with the demise of neural network research and
the continued rise of logical AI.

Many companies failed to deliver on extravagant promises.

History of AI
1969 - 1986: AI revival through expert systems
Instead of the solve-it-all optimism, researchers focused on building narrow practical systems in targeted
domains, called expert systems.

● DENDRAL: inferring molecular structure from the information provided by mass spectrometer.
● MYCIN: diagnosing blood infections.
● XCON (R1 program): the first successful commercial expert system helped configure orders for new
computer systems based on customer’s requirements.
History of AI
1987: Collapse of LISP machine and second AI winter
● Lisp machines are general-purpose computers designed to efficiently run lisp as their main software
and programming language.
● Lisp machines became expensive compared to desktop computers built by Apple and IBM, causing Lisp
machines to collapse.
● By 1990s, expert systems proved to be too expensive to maintain.
History of AI
1986 - Present: The return of neural networks
1986 Popularization of backpropagation for training multi-layer networks

1989 LeCun applied convolutional neural networks to recognize handwritten digits for USPS

1987 - Present: Probabilistic reasoning and machine learning

2001 - Present: Big data

2011 - Present: Deep learning

History of AI
2015 - 2020: AI Spring
● Unprecedented growth of research activity had revolutionary effect on most AI applications
● Availability of powerful hardware, cheap cloud computing led to this explosion of AI research

2020 - 2022: AI Winter

● Fall in the number of publications related to AI
● Reduced funding in AI research
● Probably due to the COVID pandemic

2022 - Present: AI Spring

● Advances in language translation, image recognition, game playing systems, AlexNet, ChatGPT etc.
Applications of AI
Intelligent Agents
Agents and Environments
Types of Environments
Types of Agents
Behavior and Performance Measures