Nederlands

What is AI?

Basic concepts of AI, machine learning and their applications

30 minutes

Introduction

Welcome to the first lesson of the "AI Literacy Fundamentals" module. In this lesson, we will learn what Artificial Intelligence (AI) is, how it works, and how it differs from traditional software.

The official EU AI Act definition of AI

EU AI Act - Article 3(1)

"AI system" means software that is developed with one or more of the techniques and approaches listed in Annex I and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with;

This definition is important because it determines which systems fall under the EU AI Act. Note that the definition is based on:

  • Techniques and approaches - specifically the techniques listed in Annex I of the EU AI Act
  • Functionality - the ability to generate content, make predictions, give recommendations, or make decisions
  • Impact - the influence on the environment with which the system interacts

Annex I: AI techniques and approaches

The EU AI Act specifies three categories of AI techniques in Annex I:

Machine learning

Including supervised, unsupervised and reinforcement learning, deep learning and neural networks.

Logic- and knowledge-based

Including knowledge representation, inductive (logic) programming, knowledge bases, inference engines, and deductive systems.

Statistical approaches

Including Bayesian estimation, search and optimization methods.

How AI differs from traditional software

Traditional Software

  • Follows explicit, pre-programmed rules
  • Deterministic: same input always produces same output
  • Limited to what is explicitly programmed
  • Behavior is fully predictable
  • Does not use statistical models

AI Systems

  • Learns patterns from data
  • Probabilistic: output may vary with same input
  • Can generalize to new situations
  • Behavior can sometimes be unexpected or unexplainable
  • Based on statistical models and pattern recognition

This difference is essential for understanding both the power and the risks of AI systems. Because AI systems can learn and generalize, they can perform tasks that are difficult to explicitly program, but this can also lead to unexpected behavior and difficulties in explaining decisions.

Types of AI systems and applications

Generative AI

Systems that can create new content, such as text, images, audio, or video.

Examples: ChatGPT, DALL-E, Midjourney

Predictive AI

Systems that predict future events or outcomes based on historical data.

Examples: Credit scoring, demand forecasting, fraud detection

Computer Vision

Systems that can understand, analyze, and interpret visual information.

Examples: Facial recognition, medical image analysis

Natural Language Processing

Systems that can understand, interpret, and generate human language.

Examples: Translation software, chatbots, text analysis

In your organization, you may encounter different types of AI systems, depending on your sector and functions. The EU AI Act classifies these systems based on their risk profile, which we will discuss in a later lesson.

Summary

In this lesson, we have learned:

  • The official EU AI Act definition of AI systems
  • The three main categories of AI techniques according to the EU AI Act
  • How AI differs from traditional software
  • Different types of AI systems and their applications

This knowledge forms the basis for understanding the EU AI Act and how it applies to your organization. In the next lesson, we will delve deeper into the capabilities and limitations of AI systems.