C14 - BUSI 1401 - Foundations of Information Systems - Notes on Artificial Intelligence

Chapter Outline

  • 14.1 Introduction to Artificial Intelligence

    • Explain the potential value and limitations of artificial intelligence.

  • 14.2 Machine Learning and Deep Learning

    • Differentiate among supervised, semi-supervised, unsupervised, reinforcement, and deep learning.

  • 14.3 Neural Networks

    • Describe the structure of a neural network and its purpose in machine learning.

  • 14.4 Artificial Intelligence Applications

    • Provide use case examples in computer vision, natural language processing, robotics, image recognition, and intelligent agents.

  • 14.5 Artificial Intelligence in the Functional Areas

    • Provide use case examples of AI in accounting, finance, marketing, production and operations management, human resource management, and management information systems.

14.1 Introduction to Artificial Intelligence

  • Definition of AI:

    • AI is the theory and development of information systems capable of performing tasks that typically require human intelligence.
    • Signs of intelligence include learning from experience, making sense of ambiguous messages, and responding effectively to new situations.

  • Ultimate Goal of AI:

    • To build machines that mimic human intelligence.

  • Turing Test:

    • An AI is considered intelligent if a human cannot differentiate between interacting with an AI and a human.

Types of Artificial Intelligence

  • Weak AI (Narrow AI):

    • Specialized for a single task (e.g., chatbots, voice assistants).

  • Strong AI (Artificial General Intelligence):

    • A hypothetical form of AI that can think and learn like a human across all tasks (currently not achieved).

Challenges of AI

  • Ethical Issues:

    • Job displacement.
    • Privacy concerns related to AI surveillance.
    • Emergence of AI-enabled crimes (e.g., deepfakes, misinformation, autonomous hacking).

  • Technological Advancements Contributing to AI:

    • Big Data: AI learns from enormous datasets.
    • Internet and Cloud Computing: Facilitate data access and processing.
    • Better Algorithms: Enhance efficiency and speed in learning models.
    • Advanced Hardware: Use of GPU chips for intensive AI processing.

14.2 Machine Learning and Deep Learning

  • Machine Learning (ML):

    • An AI application allowing systems to learn and improve independently from explicit programming.
    • How it Works: Systems analyze data, identify patterns, and make predictions or decisions.

  • Expert Systems (ES):

    • Systems that transfer expertise from human experts, relying on predefined rules.

Expert Systems Problems

  • Challenges in transferring human expertise to systems.
  • Difficulty in automating reasoning processes.
  • Liability issues when decisions lead to harm without human oversight.

Machine Learning vs Expert Systems

  • ESs rely on defined rules; ML learns from data.
  • ESs are not easily updated, unlike ML which thrives on more data.

Machine Learning Bias (Algorithm Bias)

  • Types of Bias:
    • Underspecification: Different models may function differently in real scenarios despite passing initial tests.
    • Developer Bias: Framing issues can affect outcomes.
    • Data Bias: Non-representative training data can lead to poor generalization.

Types of Machine Learning

  • Supervised Learning:

    • Learns from labeled input data and known outcomes.
    • Examples: Classification and regression analyses.

  • Semi-Supervised Learning:

    • Combines labeled and unlabeled data, ideal when labeled data is scarce.
    • Examples: Text classification, medical imaging.

  • Unsupervised Learning:

    • Discovers hidden patterns without pre-existing labels.
    • Example: Customer segmentation based on behavior.

  • Reinforcement Learning:

    • Learns via trial and error with rewards and penalties.
    • Applications: Self-driving cars, game-playing AIs.

  • Deep Learning:

    • Utilizes neural networks to analyze complex datasets.
    • Common uses include speech recognition and natural language processing.

14.3 Neural Networks

  • Definition:

    • Composed of nodes simulating human brain functions, enabling data learning and decision-making.

  • Structure:

    • Input Layer: Takes raw data.
    • Hidden Layer: Processes data.
    • Output Layer: Produces results.

Example - Predicting House Prices

  • Input Layer: Features like size, number of bedrooms, location.
  • Hidden Layers: Detect patterns of how features affect price.
  • Output Layer: Outputs the predicted price.

14.4 Artificial Intelligence Applications

  • Overview: AI enhances machines enabling them to assist and understand human actions.

Key Applications:

  1. Computer Vision: AI capabilities include image analysis for object detection and facial recognition.
  2. Robotics: AI integrates into robots assisting in various fields like manufacturing and healthcare.
  3. Chatbots: AI ensures 24/7 user support through interactive agents.

14.5 AI in Functional Areas

  • General Role: AI enhances business functions by improving efficiency and decision-making accuracy.

Specific Applications:

  • Accounting: Automates tasks, detects fraud (e.g., H&R Block's AI tools).
  • Finance: Employed for fraud detection, risk management, and trading automation.
  • Marketing: Enhances customer targeting and dynamic pricing strategies.
  • Production & Operations Management: Optimizes logistics and manufacturing processes.
  • Human Resource Management: Streamlines recruitment, training, and employee safety monitoring.