CS 305: Artificial Intelligence

Lecture by:

Dr. Varun Dutt

Associate Professor
School of Computing and Electrical Engineering
School of Humanities and Social Sciences
Indian Institute of Technology Mandi, India
1
 Introduction to AI

What is AI ?

AI aims to create systems that can perform tasks requiring human

intelligence.

It includes learning, reasoning, problem-solving, perception, language

understanding, and more.

What is AI? Video

2
 Overview of AI

❖ Definitions of AI:

• Thinking Humanly: Emulating human thought processes.

• Thinking Rationally: Utilizing logic and reason.

• Acting Humanly: Mimicking human behavior.

• Acting Rationally: Performing actions to achieve the best outcome.

3
 Historical Perspective of AI

Early Foundations:

Philosophers like Aristotle developed early forms of logical reasoning.

Mathematicians like George Boole created formal systems of logic.

Psychologists and linguists modeled humans and animals as

information-processing machines.

Engineers developed hardware enabling practical applications of AI.

4
 Key Milestones in AI
1940s-1950s:
- - Alan Turing proposed the concept of a universal machine.
- - The Dartmouth Conference in 1956 marked the birth of AI as a field.

1960s-1970s: Development of early AI programs like ELIZA and SHRDLU.

1980s: Expert systems like MYCIN were created for specialized tasks.

1990s: Rise of machine learning and probabilistic models.

2000s: AI integrated into web applications and everyday tools like search engines and
recommendation systems.

History of AI
5
 Historical AI Systems

ELIZA (1966): An early natural language processing computer program that

mimicked a Rogerian psychotherapist.

SHRDLU (1970): An early natural language understanding computer program

capable of interpreting and responding to commands about a
virtual world of blocks.
ELIZA & SHRDLU

- MYCIN (1976): An expert system for diagnosing bacterial infections and

- recommending treatments
MYCIN

6
 Evolution of AI

- Early AI:
- Symbolic reasoning and problem-solving programs.
- Example: Newell and Simon's GPS (General Problem Solver).

- Expert Systems:
- Rule-based systems designed to mimic the decision-making abilities
of human experts.
- Example: MYCIN for medical diagnosis.

- Machine Learning and Data-Driven AI:

- Learning from large datasets to make predictions or decisions.
- Example: Google's search algorithms.

7
 Turing Test

Concept by Alan Turing (1950):

- - A machine's ability to exhibit intelligent behavior indistinguishable from a
human.
Turing Test Video

- Test Design:
- - An interrogator converses with a machine and a human through typed
- messages.
- - If the interrogator cannot reliably distinguish the machine from the human, the
machine passes the test.

- Examples:
- - Early attempts included chatbots like ELIZA.
8
 Turing Test - Detailed Components

- Natural Language Processing (NLP): Communicating effectively in human language.

- Knowledge Representation: Storing and managing information.

- Automated Reasoning: Drawing logical conclusions from known information.

- Machine Learning: Adapting to new information and learning from experiences.

- Computer Vision and Robotics: For the total Turing Test, which includes physical
- interaction.

9
 Examples of Integration

● NLP in Practice: A machine uses NLP to understand the user’s question

about the weather.
● Knowledge Representation: The machine retrieves stored information
about the current weather.
● Automated Reasoning: The machine deduces that the user might also
want to know if they need an umbrella.
● Machine Learning: Over time, the machine learns to provide more
relevant weather-related information based on user feedback.

10
 Agents in AI
Definition: An agent is an entity that perceives its environment and takes
actions to achieve its goals.

Types of Agents:

● Simple Reflex Agents: Act based on the current perception (If condition, then
action). A thermostat that turns on the heater if the temperature drops below a
certain threshold.
● Model-Based Reflex Agents: Maintain an internal state to track the world. A
robot vacuum cleaner that keeps track of which areas have already been
cleaned.
● Goal-Based Agents: Act to achieve goals. A GPS navigation system that plans
the best route to a destination.
● Utility-Based Agents: Maximize their own perceived happiness. A self-driving
car that considers safety, speed, and comfort to choose the best route.
11
 Intelligent Agents - Components

Perception: Sensors to perceive the environment.

- Reasoning: Decision-making to determine actions.

- Action: Effectors to act in the environment.

• Examples: Autonomous vehicles use sensors, GPS, and algorithms to navigate

• and make decisions.

12
 Rational Agents

Definition: A rational agent acts to achieve the best outcome given its and capabilities.

Rationality vs Human Behavior: Rational agents aim for optimal performance

based on logic and available information, which might differ from human behavior
influenced by emotions and irrationalities.

- Examples: Chess-playing programs like Deep Blue or one for Tic-Tac-Toe use rational
strategies to win games against human champions.

13
 Modern AI Applications

Natural Language Computer Vision Robotics Healthcare

Processing
-Chatbots -Facial Recognition -Industrial Robots -Diagnostic Systems
-Language translation -Object Detection -Home assistants -Personalized
-Sentiment analysis. -Autonomous vehicles like Roomba Medicines
Finance

-Algorithmic Trading
-Fraud detection
14
 Challenges and Ethical Considerations

- Bias and Fairness:

- AI systems can perpetuate or even amplify biases present in training data.

- Transparency and Accountability:

- Understanding and explaining AI decisions.

- Privacy:
- Managing sensitive data and ensuring user privacy.

- AI Safety:
- Ensuring AI systems behave as intended and do not cause harm.

15
 Future of AI

Trends: Ethical AI Research

- Increased integration of AI in
Development: Directions:
various industries.
- Advances in reinforcement
- Creating guidelines and - Explainable AI, AI for
learning and unsupervised
regulations to ensure AI benefits social good, collaboration
learning
society as whole . between AI and humans.
- Development of more general AI
systems. 16
 Summary and Discussion

- Key Points:
- AI has a rich history and has evolved significantly over the decades.
- The Turing Test remains a fundamental, though controversial, measure of
machine intelligence.
- Intelligent agents are central to many AI applications.
- Ethical considerations are crucial in the development and deployment of AI.

- Discussion:
- What are the most significant challenges facing AI today?
- How can we ensure that AI systems are developed responsibly?

17
 References

- Books:
- "Artificial Intelligence: A Modern Approach" by Stuart Russell and Peter Norvig.

- Websites:
- AAAI (Association for the Advancement of Artificial Intelligence)
- AI Magazine and various online journals.

-Video:
- “What is AI?”--> https://youtu.be/0yCJMt9Mx9c?feature=shared
- “Evolution of AI”--> https://www.youtube.com/watch?v=lEjLs3ccap8
- “ELIZA and SHRDLU”--> https://youtu.be/DsdivS1IpeE?feature=shared
- “MYCIN”--> https://youtu.be/Yc5yP2TnqAY?feature=shared
- “Turing Test”--> https://youtu.be/3wLqsRLvV-c?feature=shared
18
 Thank you
Dr. Varun Dutt

(Email: varun@iitmandi.ac.in)

19