CSC583 - Artificial Intelligence Algorithms - Topic 1: Introduction to Artificial Intelligence

Definition of AI

  • The capability of a machine to imitate intelligent human behavior.

  • Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think and act like humans.

  • Examples

    • Siri and Alexa use AI to interpret and respond to user prompts.

Brief History

1950s

  • John McCarthy coined ”Artificial Intelligence”.

  • Alan Turing proposed the Turing Test to determine if a machine can exhibit human-like intelligence.

1960s

  • U.S. Defense Department’s investment led to the development of the first AI chatbot, ELIZA.

1970s

  • AI winter due to high expectations but limited computational power.

1980s

  • Expert systems, like MYCIN (medical diagnosis), became popular.

1990s

  • IBM’s Deep Blue defeated world chess champion, showcasing the power of AI.

2010s

  • Google’s AlphaGo defeated Go world champion, highlighting advancements in deep learning.

Alan Turing “The Father of Modern Computing”

  • Major Contributions:

    • Turing Machine: Laid the groundwork for the theory of computation and the concept of algorithms.

    • Breaking the Enigma: Played a pivotal role during WWII by devising techniques to decrypt Nazi Enigma-coded messages, significantly aiding the Allies.

    • Turing Test: Proposed a criterion for a machine to be considered ”intelligent” – if it could imitate a human to the point of being indistinguishable.

  • Legacy: Widely considered the father of theoretical computer science and artificial intelligence. His life and work have inspired numerous films, plays, and books, including the movie ”The Imitation Game.”

Turing Machine “The Foundation of Computability”

  • Definition: ”A mathematical model of computation that defines an abstract machine, which manipulates symbols on a strip of tape according to a table of rules.”

  • Components:

    • Tape: Infinite in length, divided into cells. Each cell contains a symbol from a finite alphabet.

    • Tape Head: Moves left or right one cell at a time, reads the symbol on the tape, and writes a new symbol based on the transition rules.

    • State Register: Stores the current state of the machine. The machine starts in the initial state and can transition to other states or halt.

  • Significance:

    • Universality: Turing Machines can simulate any algorithm’s logic, given enough time and tape.

    • Undecidability: There are certain problems that a Turing Machine cannot solve, leading to the concept of undecidability.

    • Foundation of Modern Computing: The concept of the Turing Machine is foundational to computer science, especially theories of computation and the limits of what computers can and cannot do.

Turing Test - Measuring Machine Intelligence

  • Definition: ”A test of a machine’s ability to exhibit intelligent behavior indistinguishable from that of a human.”

  • Concept:

    • If a human evaluator cannot reliably distinguish between the responses from a machine and a human during an interaction, the machine is said to have passed the test.

Types of AI Techniques and Methodologies

Definition

Characteristics

Use Cases

Expert Systems

Computer systems that emulate the decision-making ability of a human expert

  • Knowledge Base: Contains domain-specific knowledge.

  • Inference Engine: Applies logical rules to the knowledge base to deduce new information.

  • Medical diagnosis, financial analysis

Fuzzy Logic

A computing approach based on ’degrees of truth’ rather than the usual ’true or false’ (1 or 0) Boolean logic

  • Deals with uncertainty and vagueness.

  • Uses linguistic variables instead of numerical variables.

  • Air conditioners, washing machines, and other appliances

Artificial Neural Network (ANN)

Computing systems inspired by the structure of the human brain, consisting of interconnected nodes (neurons)

  • Can learn and make independent decisions. Used in deep learning for complex tasks

  • Image recognition, speech-to-text

Genetic Algorithm

Optimization algorithms based on the process of natural selection

  • Uses techniques such as mutation, crossover, and selection.

  • Finds approximate solutions to optimization and search problems

  • Scheduling problems, game playing

Particle Swarm Optimization (PSO)

A computational method that optimizes a problem by iteratively trying to improve candidate solutions

  • Inspired by the social behavior of birds flocking or fish schooling.

  • Each particle adjusts its position based on its experience and its neighbors.

  • Neural network training, electrical circuits

Ant Colony Optimization (ACO)

An optimization algorithm inspired by the behavior of ants in finding paths from the colony to food

  • Uses artificial ’ants’ to find good paths on a graph.

  • Pheromone updating to guide the search.

  • Routing in telecommunication networks, vehicle routing.

Machine Learning - An Overview

  • Definition: ”Machine Learning is a subset of AI where algorithms improve automatically through experience and by using data.”

  • Types of Machine Learning:

    • Supervised Learning: Algorithm is trained on labeled data.

    • Unsupervised Learning: Algorithm explores data without specific guidance.

    • Reinforcement Learning: Algorithm learns by interacting with an environment.

  • Use Cases: Routing in telecommunication networks, vehicle routing.

Definition

Key Concepts

Examples

Supervised Learning

Algorithms are trained using labeled examples, where the desired output is known

  • Input: Data that the algorithm will learn from.

  • Output: The prediction or classification made by the model.

  • Training: Process of feeding the algorithm data and allowing it to adjust.

  • Regression

  • Classification

Unsupervised Learning

Algorithms explore data without specific guidance on what to look for

  • Clusters:

  • Associations:

  • Clustering

  • Association

Reinforcement Learning

Algorithms learn by interacting with an environment and receiving feedback in the form of rewards or penalties

  • Agent: The learner or decision maker.

  • Environment: Everything the agent interacts with.

  • Actions: What the agent can do

  • Game playing

  • Robotics

Deep Learning

A subfield of machine learning that uses neural networks with three or more layers to process data in complex ways

  • Inspired by the human brain’s structure.

  • Composed of interconnected nodes or ”neurons”

  • Image / speech recognition.

  • Natural language processing.

Generative AI - Creating New Content

  • Definition: ”Generative AI refers to algorithms designed to produce or generate content, often by leveraging models like Generative Adversarial Networks (GANs).”

  • Key Concepts:

    • Training: Generative models are trained on vast amounts of data, learning patterns, structures, and features.

    • Generation: Once trained, these models can produce new, original content that mirrors the training data’s characteristics.

  • Popular Techniques:

    • Generative Adversarial Networks (GANs): Consist of two neural networks, the Generator and the Discriminator, working against each other.

    • Variational Autoencoders (VAEs): Generate new instances that can be similar to your input data.

  • Applications:

    • Art and Design: Creating artworks, music, or design elements.

    • Data Augmentation: Generating additional data for training models.

Narrow AI Specialized Intelligence

  • Definition: ”Narrow AI, also known as Weak AI, refers to artificial intelligence systems designed and trained for a specific task. Unlike General AI, which would have broad cognitive abilities, Narrow AI operates under a pre-defined set or context.”

  • Characteristics:

    • Task-Specific: Designed for a single, narrow task.

    • No Consciousness: Operates based on data and algorithms, without feelings, desires, or consciousness.

    • Prevalent Today: Most current AI applications are forms of Narrow AI.

  • Examples:

    • Voice Assistants: Siri, Alexa, and Google Assistant.

    • Recommendation Systems: Netflix movie suggestions, Spotify music playlists.

    • Image Recognition: Software that identifies objects in photos.

General AI Broad Cognitive Abilities

  • Definition: ”General AI, also known as Strong AI, refers to a form of artificial intelligence that has the ability to understand, learn, and perform any intellectual task that a human being can.”

  • Characteristics:

    • Versatile Learning: Can learn and perform multiple tasks, not restricted to one domain.

    • Human-like Cognition: Processes information and makes decisions in a manner similar to human reasoning.

    • Theoretical: Currently, General AI remains a concept and has not been realized.

  • Examples:

    • Complexity: Replicating the vast range of human cognitive abilities.

    • Ethical Concerns: Implications of creating machines with human-like consciousness.

Narrow Al vs General Al - Understanding the differences

Aspect

Narrow Al (Weak AI)

General Al (Strong AI)

Definition

Specialized for a specific task.

Capable of any intellectual task that humans can do.

Examples

Chatbots, image recognition software, voice assistants

Theoretical robots or systems that can learn and adapt to any environment or task

Current State

Prevalent in today's technology

Remains a concept; not yet realized

Scope

Task-specific

Broad capabilities

Learning

Operates within its training

Can learn and adapt outside its initial programming

Existence

Widespread in current applications

Still theoretical

Ethics in AI Navigating the Moral Landscape

  • Key Ethical Concerns:

    • Bias and Fairness: AI models can inherit and amplify biases present in training data, leading to unfair decisions.

    • Transparency and Accountability: Understanding how AI makes decisions (explainability) and who is responsible when AI goes wrong.

    • Privacy: Ensuring user data is protected and not misused.

    • Safety and Security: Ensuring AI systems operate safely and are resistant to malicious attacks.

  • Challenges:

    • Balancing innovation with ethical considerations.

    • Establishing global standards and regulations.

Social Impact of AI - Changing the Fabric of Society

  • Positive Impacts:

    • Efficiency and Productivity: Automating tasks can lead to increased productivity in various sectors.

    • Innovation: New AI-driven technologies and solutions can address complex societal challenges.

    • Personalization: Tailored experiences in areas like education, healthcare, and entertainment.

  • Concerns:

    • Job Displacement: Automation might replace certain jobs, leading to unemployment in specific sectors.

    • Surveillance and Privacy: Potential misuse of AI in surveillance and erosion of personal privacy.

    • Social Manipulation: AI-driven fake news, deepfakes, and targeted advertising can influence public opinion.