Artificial Intelligence Fundamentals

Introduction to Artificial Intelligence (AI)

• AI's birth was in 1956.
• AI is a subfield of computer science concerned with how computer programs/machines can think and act like humans.
• AI simulates human intelligence in machines, programming them to think like humans and mimic their actions.
• AI automates activities associated with human thinking, including decision-making, learning, planning, reasoning, searching, acting, and problem-solving.
• It aims to create computer programs or machines to perform functions requiring intelligence when done by people.
• AI focuses on automating intelligent behavior based on computational models.

Goals of AI

• Artificial: Created by human effort, not naturally occurring.
• Intelligence: The ability to acquire knowledge and use it (Pigford and Baur).
• AI goals: Making computers more helpful by taking on dangerous or tedious tasks from humans, and understanding the principles of human intelligence.

Organization of AI Definitions

• Categorized into four groups:
  • Systems that think like humans
  • Systems that think rationally
  • Systems that act like humans
  • Systems that act rationally

What is Artificial Intelligence (AI)?

• Making computer programs think and act like humans (rationally and intelligently).

Systems that think like humans: Cognitive Modeling

• Cognitive modeling: Process of building computer-based simulations/models to mimic how humans think and process information.
• Observing humans from the inside to understand how they think (introspection vs. psychological experiments).
• AI goal: Mimic human minds (Haugeland)
• Activities like decision-making, problem-solving, and learning are automated. (Bellman)

What about the brain?

• Brains excel at making rational decisions(though not perfect).
• Brains are analogous to software wings in flight, not modular like software.
• Prediction and simulation are key to decision-making.

Systems that think rationally: "Laws of Thought"

• Humans are not always rational.
• Rationality is defined in terms of logic, but logic cannot encompass everything (e.g., uncertainty).
• Logical approaches can be computationally expensive.
• Thinking rationally is understanding the mental faculties through computational models.(Charniak and McDermott)
• Thinking rationally is understanding computations enabling perception, reasoning, and action.(Winston)

Turing Test

• The Turing Test approach: A human questioner cannot differentiate between a computer and a human based on responses.
• Intelligent behavior: Human-level performance in all cognitive tasks.
• Total Turing Machine: Includes computer vision and robotics to perceive and move objects, respectively.

Cognitive Tasks

• Natural Language Processing: For human communication.
• Knowledge Representation: Effective and efficient storage of information.
• Automated Reasoning: Retrieving and answering questions based on stored information.
• Machine Learning: Adapting to new situations.

Artificial Intelligence (AI) Activities/Tasks/Goals

• Searching
• Planning (using logic)
• Knowledge representation (using logic)
• Reasoning (using logic)
• Prediction
• Learning
• Decision-making (using logic)
• Problem-solving (using logic)
• Perception (using logic)
• Simulation

Artificial Intelligence (AI) Areas/Applications

• Machine learning
• Deep learning
• Natural language processing (NLP)
• Computer vision
• Image processing
• Internet of things
• Robotics
• Cyber security
• Virtual reality
• Social media
• Agriculture
• Business, marketing, and finance
• Industry/manufacturing
• Sports
• Healthcare
• Transportation
• Expert systems
• Email spam filtering
• Climate forecasting
• Entertainment
• Engineering
• Software engineering
• Recommendation systems
• Game playing (e.g., chess)
• Bioinformatics
• Politics and intelligence
• Music composition
• Web search engines
• Education
• Government (Surveillance)
• E-commerce
• Technology and smart devices
• Travel and logistics

Natural Language Processing Applications

• Chatbots
• Sentiment analysis
• Machine translation
• Information retrieval
• Question answering
• Speech recognition
• Email filtering

Some of the Main Activities and Areas of Al and Their Interdependencies

• A diagram showing the interconnections between various AI activities (Search, Logic, Knowledge Representation, Machine Learning, Planning, NLP, Vision, Robotics, and Expert Systems).

Systems that Act Rationally: "Rational Agent"

• Rational behavior: Doing the correct thing.
• The right thing: Maximizing goal achievement given available information.
• A rational agent does not need to replicate human thought processes or make the same decisions as humans.
• Uses purely logical reasoning instead if necessary.

Rational Agents

• An agent is an entity that perceives and acts.
• Agents function as a function of perceived histories and taking actions following it.
• Seek the agent with the best performance, considering environment and given tasks.
• Unattainable perfect rationality due to computational limits.
• Design programs that best utilize available machine resources.

Search

• Search is the basic technique in AI.
• Possible answers, decisions, and actions are structured as an abstract space, and searched through in order to find solutions.
• Search can be categorized as uninformed, or informed(with prior assumptions).

Knowledge Representation & Reasoning

• Knowledge representation and reasoning are crucial for rational behavior in environments.
• Environments need to be described, or represented, allowing for inferences to be made.
• Methods used for describing the world effectively: How to describe it concisely? How to get hold of the right information when needed? How do we generate new knowledge? How do we deal with uncertainty?

Advantages of Artificial Intelligence

• More powerful and useful computers.
• Improved user interfaces.
• Solution to new problems
• Better information handling
• Relieved information overload.
• Conversion of information to knowledge

Disadvantages of Artificial Intelligence

• High cost
• Difficult and costly to develop software.
• Few experienced AI programmers.
• Little practical AI products that are implemented in the market.

Machine Learning

• Machine learning (ML) is a subfield of artificial intelligence (AI) focused on computer programs improving performance through experience using data.

Life Cycle of Machine Learning

• Input data (training dataset)
• Machine Learning
• Testing dataset
• Output training model
• Output
• Predictions, Classification, Clustering, Association, Learning

Types of learning using Machine Learning

• Supervised learning: Learning from labelled data.
• Unsupervised learning: Learning from unlabeled data.
• Semi-supervised learning: Learning from labeled and unlabeled data.
• Reinforcement learning: Learning based on feedback.

Machine Learning Tasks

• Regression (Prediction)
• Classification
• Clustering
• Association
• Learning

Test your understanding of the key concepts and definitions related to artificial intelligence. This quiz covers the primary functions, goals, and activities associated with AI, as well as cognitive modeling. Challenge yourself to distinguish between accurate and inaccurate statements about AI.