Artificial Intelligence

Bipun Man Pati, D.Eng.

Introduction (My Background)
2014 Masters Degree (M.Eng.,)
Telecommunication Engineering
Thailand

2017 Doctoral Degree (D.Eng.,)

Telecommunication and Artificial
Intelligence
Thailand, Japan, USA

2022 PostDoc
Artificial Intelligence
Thailand, Japan

2023 Chief, AI Research Center

Advanced College of Engineering and
Management
Kathmandu, Nepal
Research Profile- Google Scholar
Research Interests-
● Wireless
Communication
● Artificial Intelligence
● Telecommunications
Course Contents- Theory
1. Introduction to Artificial Intelligence (4 hrs)
2. Intelligent Agents (5 hrs)
3. Problem Solving and Search Algorithms (10 hrs)
4. Knowledge Representation and Reasoning (10 hrs)
5. Machine Learning (6 hrs)
6. Fuzzy Logic (5 hrs)
7. Expert System (5 hrs)
Course Contents- Practical Work
1. Implement search algorithms (e.g., BFS, DFS, A*)
2. a) Create a program with facts about a different domain (e.g., a book collection, family tree) and
perform queries to extract information.
b) Extend the family tree with new rules to find relationships between grandparents, siblings, or
cousins. Enhance the program with additional rules and variables to solve a given set of queries.
Include comments explaining the logic.
c) Develop a program that solves a specific problem related to a domain of interest (e.g., a
recommendation system) and prepare a brief report explaining the design and logic.
d) Implement Semantic Network in for Knowledge Representation and Querying (For example, a
simple hierarchy might involve concepts such as “Animal”, “Bird”, and “Mammal” with relationships
such as is_a, has_parts, or can_fly)
e) Implement a Frame-Based Representation with Inheritance and Queries.
3. Develop a simple expert system using rule-based reasoning/fuzzy logic.
4. Implement and evaluate classification algorithms (e.g. linear regression and k-NN)
5. Build and train single layer and multi-layer perceptron.
6. Implement the fuzzy logic for reasoning in an expert system.
Course Requirements & Prereqs
Uses a variety of skills / knowledge:

• Probability and statistics

• Boolean Logic

• Algorithms

• Above average coding skills

Official Prerequisites– Data structures

Text Book
1. Russell, S. J., & Norvig, P. (2022). Artificial intelligence: a modern
approach. Pearson.
Chapter 1
1.1. Intelligence

1.1.1. Types of Intelligence

1.1.2. Components of Intelligence

1.2. Artificial Intelligence

1.2.1. Approaches of AI

1.2.1.1. Acting Humanly

1.2.1.2. Thinking Humanly

1.2.1.3. Thinking Rationally

1.2.1.4. Acting Rationally

1.2.2. Foundations of AI

1.2.3. History of AI

1.2.4. Risk and Benefits of AI

1.3. Ethics and Societal Implications

1.3.1. Ethical Implications of AI

1.3.2. AI and Society: Work and Automation, Employment, Privacy and Security

1.3.3. Governance and Regulation

History of AI
1946: ENIAC heralds the dawn of
Computing
● first programmable,
electronic, general-purpose
digital computer
● Turing complete
● Able to solve "a large class of
numerical problems" through
reprogramming
What will i use computer in 40’s?
- Artillery firing tables for USA
- Human computer 20 hrs
- ENIAC 30 secs
1950: Turing asks the question…
● Father of Computer Science
● I propose to consider the
question: “Can machines
think?” -Alan Turing, 1950
● Brain is meaningless in the
absence of the communicators
in the body
1956: A new field is born
● A 2 month, 10 man study of artificial
intelligence be carried out during the
summer of 1956 at Dartmouth College in
Hanover, New Hampshire.
● Dartmouth AI Project Proposal; J. McCarthy
et al.; Aug. 31, 1955.
● Founding fathers of AI
● First time word “AI” is used
● In 2 months we will solve AI
● We have not still not solved
● AI researchers are highly optimistic
1964 - 1966
1964: Eliza - The first chatbot psychotherapist
● One person started chatting with
Eliza and Eliza started asking
personal question and this person
started answering personal question
and felt there was a real
psychotherapist sitting on the other
side
1966: Shakey - First General purpose mobile
robot (Stanford)
● The most important algorithm in AI-
A* search algorithm
AI Winter
Things started to go bad for AI

1974-1980: first winter

● Failure of machine translation (Eng-> Russian)

● Negative results in Neural Networks (XOR)
● Poor speech understanding

Until now not solved fully

1987-1993: second winter

● Expert system era- 90’s (old style AI course)

● Idea was somebody uses logical language to encode all knowledge about their field automated
doctor
● Hardware for expert system
● New machines came out and the h/w for expert system declined
AI Winters
AI researchers were very optimistic Lasting effects
They kept making false claims
– [Economist 07] “Artificial Intelligence
AI started getting bad names
is associated with systems that have all
People get stuck in time
too often failed to live up to their
People started to cut off the cord with AI, there promises.”
were topics that were part of AI but created their
own sub communities they started saying
– [Pittsburgh BT06] “Some believe the
● we are not AI we are CV word 'robotics' actually carries a stigma
● we are not AI we are NLP
● we are not AI we are ML that hurts a company's chances at
funding.”
AI in the 90’s
1997: Deep Blue (IBM) ends
Human Supremacy in Chess

“I could feel human-level

intelligence across the room”
-Gary Kasparov, World Chess
Champion (human)
In a few years, even a single
victory in a long series of games
would be the triumph of human
genius.
On May 11, 1997, it won a 6-game match by 2 wins to 1 with
3 draws
Success story in AI
● Does Deep Blue use AI? or just a search system and choosing the best path

“If it works it is not AI

if it works people say it's just an algorithm

● Magical door back 120 years

just a motion sensor

● No more magic
● AI is the next thing that does not work

Saying Deep Blue doesn’t really think about chess is like saying an airplane doesn’t really fly because it
doesn’t flap its wings.

– Drew McDermott
2004 & 2009
● Algorithm for the control of robot
on mars
○ Small ditch stop “I am expensive”
○ Waiting question and answer

Planner was devised by AI research

working at NASA at that time

Operation people's ”How do I trust

this AI system”
2005 DARPA Grand Challenge
● AI still did not touch general public
● Long trip taken by a car itself
● Stanley and three other cars drive
themselves over a 132 mile mountain
road in Navada desert.
● Teams
○ Stanford University (Profesor Sebastian -
online AI course with Peter Norvig) -160,000
peoples. Worked with undergrads students-
Stanley
○ CMU - professor Whitaker - very famous for
self driving cars for long history https://youtu.be/LZ3bbHTsOL4?si=T
■ Two entry
● Highlander 6OSQPMixzy6idJl
● Sandstorm- just follow leader and
come second - 2nd position
2007 DARPA Urban Challenge
● Whitaker made new car “Tartan“ and won the DARPA urban challenge.
● https://youtu.be/lULl63ERek0?si=yBOfka6eyvwep_YL
● 12 year anniversary- 2017 got the car back from garage and cleaned up and
find out the problem- just one component went wrong and they lost the
challenge.
2011: IBM’s Watson
IBM’s next big machine IBM Watson, a
jeopardy player defeated Ken Jennings
(World champion in jeopardy)

https://www.youtube.com/watch?v=Sp4
q60BsHoY
AI: Present after 2011
● Interesting new developments in AI after 2011 causing AI as a field to change
● History of AI in terms of Application but not technology
● Going back in terms of technology initially there was a lot of work on search
and also representation of Neural network for representing functions

Logic based AI, Search AI (before 90’s)

Probabilistic AI (90’s to 2010)→ Bayesian Network

Neural Network

One of the prominent Demonstration in AI was AlphaGo

Present state of AI
2016: AlphaGo (Google DeepMind)
● AlphaGo mastered the ancient
game of Go, defeated a Go
world champion (Lee Sedol),
and inspired a new era of AI
systems.
What has Changed now?
Algorithms that we still use to do AI are
quite old

Compute has become very important

Object Recognition
Artistic Applications
Style transfer

Image Colorization

Image Super Resolution

Image Caption
Automatic Speech Recognition
Where is AI?
Huge progress in AI in the past decade

”If it works it is not AI” —> “It’s all AI”

Lot of funding in AI “If you want funding call it AI”

Speed of progress —> lots of hype

What is being said about AI?

A recent survey of 50 Nobel Laureates ranked the climate, population rise, nuclear war, disease,
selfishness, ignorance, terrorism, fundamentalism, and Trump as bigger threats to humanity than AI.
What is being said about AI?

It is therefore better to understand what it is able to do and what not to be much prepared.
RoboCup 2050- IEEE Robotics & Automation Magazine

The ultimate goal of the RoboCup initiative is stated as

follows [1]:
“By mid-21st century, a team of fully autonomous
humanoid robot soccer players shall win the soccer
game,comply with the official rule of the FIFA, against the
winner of the most recent World Cup.”

https://youtu.be/pCLD93Uuzac?si=fvNhCf8M8OLQrETl

https://www.researchgate.net/publication/3344615_The_road_to_RoboCup_2050
What is AI?
What is AI?
AI is fairly a controversial subject

AI from science perspective

Artificial intelligence: What is the nature of intelligent thought?

What is intelligence?
Dictionary.com: capacity for learning, reasoning, understanding, and similar forms
of mental activity

● Ability to perceive and act in the world

● Reasoning: proving theorems, medical diagnosis
● Planning: take decisions
● Learning and Adaptation: recommend movies, learn traffic patterns
● Understanding: text, speech, visual scene
Human and Intelligence
● Are humans intelligent?
○ Replicating human behaviour has been traditionally been considered as
hallmark of intelligence.
● Are humans always intelligent?
we need to disassociate the definition of intelligence to what human do.
● Can non-human behavior be intelligent?
○ We hear news dogs can hear certain frequency that human cannot.
○ There is intelligence out there which is non-human

What was considered intelligent before is not considered intelligent now.

Conclusion: Intelligence is more than just what humans do
 What is Artificial Intelligence
● Historically, researchers have pursued several different versions of AI.
Approaches to Artificial Intelligence

This approach is what

most people agree
Thinking Humanly: The cognitive modeling approach
● This approach had few takers and also simple to critique.
● Thinking is innate process it comes naturally.
● Cognitive Science
○ Very hard to understand how humans think
○ Post-facto rationalizations, irrationality of human thinking
● Do we want a machine that beats humans in chess or a machine that thinks like
humans while beating humans in chess?
○ Deep Blue supposedly DOESN’T think like humans.
● Human have hardware of brain, the machine do not have the same hardware as
humans. So it is hard to claim that we can build machine that can think like human
● Intelligence is more than human and thinking is not just sufficient.
The goal of aeronautical engineering is not to fool pigeons in flying!!!
Thinking Rationally: Laws of thought approach
● Lot of people believe this is good approach to AI.
● Aristotle: what are correct arguments/thought processes?
○ Logical deliberation is important for intelligent behavior.
● Problems
○ Not all intelligent behavior is mediated by logical deliberation (reflexes)
○ What is the purpose of thinking rationally?
■ Should the fact that i can think is sufficient for intelligence or should i be focused my
thinking in certain direction.
Acting Humanly: The Turing test approach

If the human cannot tell whether the responses from the other side of a wall are coming from a human
or computer, then the computer is intelligent.
Acting Humanly: The Turing test approach
● Loebner Prize
○ Every year in Boston
○ Expertise-dependent tests: limited conversation
● What if people call a human a machine?
○ Shakespeare expert
○ Machine did not win truing test, the human failed in the turing test.
○ Make human-like errors to win turing test.
○ Do we want to make machine that makes human like error so that it can win turing test. Hence
this is not the correct approach to AI.
● Problems
○ Not reproducible, constructive or mathematically analyzable
Acting rationally: The Rational Agent Approach
● This is the most appropriate approach to AI.
● Rational behavior: doing the right thing
● Need not always be deliberative
○ Reflexive
● Aristotle (Nicomachean ethics)
○ Every art and every inquiry, and similarly every action and every pursuit is thought to aim at
some good.
● Problem:
○ What is rational?
Acting vs thinking rationally
● Weak vs Strong AI Hypothesis
○ Weak Hypothesis: Machine can act intelligently without thinking intelligently.
■ Example: Convert English to Chinese
■ Huge book with every possible sentence from english to chinese.
■ When a query comes i go and see the book and sends out the sentence.
■ Does that person knows both language? Yes!
■ Can that person fool other person? No!
○ Strong Hypothesis: Machine that act intelligently should also think intelligently too.
● Some people Acting rationally is also not the good approach to AI as you can
act rationally with or without thinking rationally.
 The Foundations of AI

● A brief history of the disciplines that contributed ideas, viewpoints, and

techniques to AI.
● We organize the foundation of AI around a series of questions from various
disciplines
○ Philosophy
○ Mathematics
○ Economics
○ Neuroscience
○ Psychology
○ Computer Engineering
○ Control theory and Cybernetics
○ Linguistics
The Foundations of AI- Philosophy

● Can formal rules be used to draw valid conclusions?

● How does the mind arise from a physical brain?
● Where does knowledge come from?
● How does knowledge lead to action?
The Foundations of AI- Mathematics

● What are the formal rules to draw valid conclusions?

● What can be computed?
● How do we reason with uncertain information?
The Foundations of AI- Economics

● How should we make decisions in accordance with our preferences?

● How should we do this when others may not go along?
● How should we do this when the payoff may be far in the future?
The Foundations of AI- Neuroscience

Neuroscience is the study of the nervous system, particularly the brain.

● How do brains process information?

The Foundations of AI- Psychology

● How do humans and animals think and act?

The Foundations of AI- Computer engineering

● How can we build an efficient computer?

The Foundations of AI- Control theory and cybernetics

● How can artifacts operate under their own control?

The Foundations of AI- Linguistics

● How does language relate to thought?

 Risks and Benefits of AI

Risks from the misuse of AI.

Some of the risks are already apparent,
while others seem likely based on current
trends:
● Lethal Autonomous
Weapons(LAWs):
○ These are defined by the United
Nations as weapons that can locate,
select, and eliminate human targets
without human intervention.
○ The technologies needed for
autonomous weapons are similar to
those needed for self-driving cars.

https://youtu.be/8GwBTFRFlzA?si=C1OVhJ1_P8Aeubd2
 Risks and Benefits of AI (Contd …)

● Surveillance and Persuasion:

○ It is expensive, tedious, and
sometimes legally questionable for
security personnel to monitor phone
lines, video camera feeds, emails,
and other messaging channels
○ AI can be used in a scalable fashion
to perform mass surveillance of
individuals and detect activities of
interest.
○ By tailoring information flows to
individuals through social media,
based on machine learning
techniques, political behavior can be
modified and controlled to some
extent
 Risks and Benefits of AI (Contd …)

● Biased Decision Making

○ Careless or deliberate misuse of machine learning algorithms for tasks such as
evaluating parole and loan applications can result in decisions that are biased by race,
gender, or other protected categories.
 Risks and Benefits of AI (Contd …)

● Impact on employment
○ Concerns about machines eliminating jobs are centuries old.
○ Previous advances in technology—such as the invention of mechanical looms—have
resulted in serious disruptions to employment, but eventually people find new kinds of
work to do.
○ On the other hand, it is possible that AI will be doing those new kinds of work too.
○ Example: Self-driving cars
 Risks and Benefits of AI (Contd …)

● Safety critical applications

○ As AI techniques advance, they are
increasingly used in safety-critical
applications such as driving cars and
managing the water supplies of cities.
○ Fatal accidents have already occurred.

https://youtu.be/6Kf3I_OyDlI?si=apxsGe4zVhaR0yUH
 Risks and Benefits of AI (Contd …)

● Cybersecurity
○ AI techniques are useful in defending against cyber attack.
○ For example, reinforcement learning methods have been used to create highly effective
tools for automated phishing attacks.
 Risks and Benefits of AI (Contd …)

Benefits
● Our entire civilization is the product of our human intelligence.
● If we have access to substantially greater machine intelligence, the ceiling on our ambitions is raised
substantially.
● The potential for AI and robotics to free humanity from menial repetitive work and to dramatically
increase the production of goods and services could presage an era of peace and plenty.
● The capacity to accelerate scientific research could result in cures for disease and solutions for
climate change and resource shortages.

Demis Hassabis, CEO of Google DeepMind, has suggested: “First solve AI, then use AI to solve
everything else.”
Applications of AI

● Autonomous Planning and Scheduling

● Game Playing
● Autonomous Control
● Diagnosis
● Logistics Planning
● Robotics
● Language understanding and Problem Solving
 Risks and Benefits of AI (Contd …)

Benefits
● Our entire civilization is the product of our human intelligence.
● If we have access to substantially greater machine intelligence, the ceiling on our ambitions is raised
substantially.
● The potential for AI and robotics to free humanity from menial repetitive work and to dramatically
increase the production of goods and services could presage an era of peace and plenty.
● The capacity to accelerate scientific research could result in cures for disease and solutions for
climate change and resource shortages.

Demis Hassabis, CEO of Google DeepMind, has suggested: “First solve AI, then use AI to solve
everything else.”
Question
Define artificial Intelligence. Explain various application of Artificial intelligence in
real field. What are the challenges of Artificial Intelligence?

Define artificial intelligence? Discuss the approaches of Al.