Back

01_Introduction

Loading views...
Next →

Artificial Intelligence - Introduction

What is Artificial Intelligence (AI)?

Simple Definition

Artificial Intelligence (AI) is the science of making computers and machines smart enough to perform tasks that normally require human intelligence.

In Simple Terms:

AI enables machines to:

  • Learn from experience (like humans learn from mistakes)
  • Understand language (reading, writing, speaking)
  • Recognize patterns (faces, objects, sounds)
  • Make decisions (choosing the best option)
  • Solve problems (finding solutions to complex challenges)

Everyday Examples:

  • Siri/Alexa - Understanding your voice commands
  • Netflix/YouTube - Recommending videos you might like
  • Google Maps - Finding the fastest route
  • Face Unlock - Recognizing your face on your phone
  • Spam Filters - Detecting unwanted emails
  • Auto-Correct - Fixing your typing mistakes

Four Aspects of AI Definition (Academic View)

We define AI in 4 aspects based on two dimensions:

  • Human vs. Rational approach
  • Thinking vs. Acting behavior

1. Thinking Humanly

Systems that think like humans - the <abbr title="Studying how the human brain thinks and copying it in computers">Cognitive Modeling</abbr> approach.

  • <abbr title="Looking inside your own mind to understand how you think">Introspection</abbr>: Catching our own thoughts as they go by.
  • Psychological Experiments: Observing a person in action.
  • Brain Imaging: Observing the brain in action (fMRI, EEG).
  • Cognitive Science: The interdisciplinary field that brings together computer models from AI and experimental techniques from psychology to construct precise and testable theories of the human mind.
  • Example: <abbr title="Computer systems designed like the human brain with connected nodes">Neural networks</abbr> inspired by human brain

2. Acting Humanly

Systems that act like humans - the Turing Test approach.

  • Focuses on behavior that appears human.
  • Requires capabilities like Natural Language Processing, Knowledge Representation, Automated Reasoning, and Machine Learning.
  • Total Turing Test: Includes a video signal and physical hatch for interaction (requires Computer Vision and Robotics).
  • Example: Chatbots, virtual assistants

3. Thinking Rationally

Systems that think rationally - the "Laws of Thought" approach.

  • Syllogisms: Aristotle's patterns for argument structures that always yield correct conclusions when given correct premises (e.g., "Socrates is a man; all men are mortal; therefore, Socrates is mortal").
  • Logicist: The tradition in AI that hopes to build on such programs to create intelligent systems.
  • Example: <abbr title="Computer programs that solve problems like human experts do">Expert systems</abbr>, <abbr title="Programs that prove mathematical statements are true or false">theorem provers</abbr>

4. Acting Rationally

Systems that act rationally - the Rational Agent approach.

  • Rational Agent: One that acts so as to achieve the best outcome or, when there is uncertainty, the best expected outcome.
  • Standard Model: The pursuit of the right action (rationality) as the primary goal of AI.
  • Limited Rationality: Acting appropriately when there is not enough time to do all the computations one might like.
  • Example: <abbr title="Able to work independently without human control">Autonomous</abbr> vehicles, game-playing AI

State of the Art

What can AI do today? While we are far from general human-level intelligence, AI excels in specific domains:

  • Robotic Vehicles: Autonomous cars (Waymo), Mars rovers.
  • Machine Translation: Translating between hundreds of languages (Google Translate).
  • Speech Recognition: Voice assistants and transcription.
  • Recommendations: Content discovery on YouTube, Amazon, and Netflix.
  • Game Playing: Superhuman performance in Chess, Go (AlphaGo), and StarCraft II.
  • Image Understanding: Medical diagnosis from X-rays, facial recognition.
  • Medicine: Drug discovery, surgical robots.
  • Climate Science: Predicting weather patterns and mapping deforestation.

The Value Alignment Problem

As AI systems become more capable, it is critical that they are value-aligned—their goals must match human intentions. If a system is perfectly rational but pursues the wrong goal (e.g., "eliminate cancer" by "eliminating all humans"), it is a failure of Value Alignment.

Summary

AI is about creating intelligent machines that can:

  • Think and act like humans (human-centered approach)
  • Think and act rationally (rationality-centered approach)

The field combines computer science, mathematics, psychology, linguistics, and philosophy to build systems that exhibit intelligent behavior.

Next: Turing Test →

Next →