Cognitive Science and AI Applications

Introduction to AI

• AI's birth year: 1956
• Definition: A subfield of computer science focused on creating computer programs/machines that can think and act like humans.
• AI aims to simulate human intelligence in machines.
• Key Activities: Automation of human activities like decision-making, learning, planning, reasoning, searching, acting, and problem-solving.
• AI is used to create computer programs/machines that perform functions requiring intelligence when handled by humans.
• AI is concerned with automating intelligent behavior using computational models.

Goals of Artificial Intelligence

• Artificial: Created by humans; not occurring naturally.
• Intelligence: Ability to acquire knowledge and apply it (Pigford and Baur).
• Goals: Making computers more useful by taking on tedious or dangerous tasks from humans; Understanding the principles of human intelligence.

Organization of AI Definitions

• Categories: Four categories based on thought vs behaviour (human vs rational).
  • 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 modelling

• Cognitive Modeling: Creating computer-based simulations to imitate human thought processes and information processing.
• How humans think: Research methods include introspection and psychological experiments.
• Cognitive Science: The effort to make computers capable of genuine thinking (Haugeland).
• Activities associated with human thinking: Decision-making, problem-solving, learning (Bellman).

What about the brain?

• Human brains excel at making rational decisions, but aren't perfect.
• Brains act as intelligent guides similar to the way wings function to control flight.
• Brains aren’t as modular as software.
• Prediction and simulation are vital to decision-making, derived from learned lessons.

Systems that think rationally ("laws of thought")

• Humans are not always rational.
• Rationality is defined in terms of logic, but logic cannot completely account for factors such as uncertainty.
• Logical approaches may not always be feasible from a time, or computation perspective.
• Thinking rationally is studying mental faculties using computational models (Charniak and McDermott).
• Thinking rationally is concerned with the algorithms that enable thinking and acting (Winston).

Turing Test

• Approach: Determining whether a computer can exhibit intelligent behaviour by imitating human conversation.
• Intelligent Behavior: Achieving human-level performance in all cognitive aspects.
• Turing Machine Considerations: Includes computer vision and robotics aspects.
• Computer Vision: Perceiving objects (seeing).
• Robotics: Moving objects (acting).

Cognitive Tasks

• Natural Language Processing: Used for communication between humans and computers.
• Knowledge Representation: Storing information effectively and efficiently.
• Automated Reasoning: Retrieving and answering questions using stored information.
• Machine Learning: Adapting to new circumstances.

Artificial Intelligence (AI) Activities/Tasks/Goals

• Activities: Examples include searching, planning, knowledge representation, reasoning, prediction, learning, decision-making, problem-solving, perception, and simulation.

Artificial Intelligence (AI) Areas/Applications

• Areas/Applications: Examples include machine learning, deep learning, natural language processing, computer vision, image processing, internet of things, robotics, cyber security, virtual reality, and social media.
• Further areas/applications: agriculture, finance, industry, sports, healthcare, transportation, experts systems, email filtering, climate forecasting, entertainment, engineering, software engineering, recommendation systems, game playing, bioinformatics, politics, music composition, web search engines, education, government, e-commerce, technology, and travel/logistics

Natural Language Processing Applications

• Examples: Chatbots, sentiment analysis, machine translation, information retrieval, question answering, speech recognition, and email filtering.

Some of the Main Activities and Areas of AI and their Inter-dependencies

• Graph: Shows the relationships between activities and areas of AI, illustrating how they interact.

Systems that Act Rationally: "Rational Agent"

• Rational Behavior: The act of doing the right thing. This means maximizing goal achievement given the information available.
• Rational Agent Requirements: The agent should replicate human thought processes, make the same decisions as humans, and use purely logical reasoning.

Rational Agents

• Definition: Entitites that perceive and act.
• Function: A function mapping perception history (P*) to an action (A).
• Goals: Finding the best possible agent for a given type of environment and task.
• Limitations: Computational limitations can make perfect rationality unattainable.

Search

• Fundamental AI Technique: Organizing potential answers, decisions into an abstract search space.
• Search Types: Blind (moves through the space without considering future implications) and Informed (uses information to guide the search process).

Knowledge Representation & Reasoning

• Core AI Concept: Needed to describe complex situations and make use of knowledge logically.
• Crucial for describing the environment, and drawing inferences.
• Components involved: Describing the world and knowing how to describe it concisely, describing situations in a way that can generate correct information and deal with uncertain information.

Some Advantages of Artificial Intelligence

• Increased computing power
• Improved interfaces
• Solving novel problems
• Better information handling
• Reduced information overload
• Converting data to actionable knowledge

The Disadvantages of Artificial Intelligence

• Increased costs
• Software development challenges (slow and expensive)
• Limited experienced programmers
• Lack of practical products available on the market

Machine Learning

• Subfield of AI: Focuses on computer programs that enhance their performance through experience and data use.

Life Cycle of Machine Learning

• Process steps: Input data, machine learning, output (training model), testing dataset, classification, predictions, clustering, associations, and learning.

Types of Learning using Machine Learning

• Supervised learning: Learning from labelled data.
• Unsupervised learning: Learning from unlabeled data.
• Semi-supervised learning: Learning from a mix of labeled and unlabeled data.
• Reinforcement learning: Learning by receiving feedback from an environment.

Machine Learning Tasks

• Types of Tasks: Regression (prediction), classification, clustering, association, and learning itself.

Explore the fascinating intersection of cognitive science and artificial intelligence with this quiz. Test your understanding of key concepts, methods, and applications in both fields. Delve into the theories and technologies that define human thinking and machine learning.