Chapter 1. Artificial Intelligence in Everyday Life

1.1 What is artificial intelligence?

Artificial intelligence (AI) is a broad term for computer-based systems designed to perform tasks that commonly require human-like capabilities, such as recognising patterns, interpreting language, generating content, making predictions, planning, or supporting decisions. There is no single technique called AI. The field includes statistical models, machine learning, deep learning, language technologies, computer vision, robotics, and systems that combine several methods.

AI should be understood as a capability within a larger social and technical system. A model may produce a prediction, but people decide what information to provide, how the result is used, what safeguards are needed, and who is accountable. A useful AI service therefore includes data, software, interfaces, operating procedures, monitoring, and human oversight—not just a model.

1.2 Where people encounter AI

Common examples include search ranking, spam filtering, translation, speech-to-text, route planning, recommendation systems, fraud alerts, document classification, image enhancement, accessibility tools, and conversational assistants. Some systems operate in the background and make a prediction; others generate a response directly for a user.

Generative AI creates new content such as text, images, audio, video, or code in response to instructions or other inputs. The OECD describes generative AI as a rapidly adopted family of technologies with applications across education, creative work, healthcare, and software development, while also identifying risks involving bias, manipulated media, and disinformation (OECD, n.d.-a).

1.3 Why AI matters now

AI matters because it can reduce the time needed for some information-intensive tasks, make interfaces more accessible, help people explore large datasets, and enable new products and services. It can also shift the cost and speed of producing convincing content, including false content. The same capability that helps draft a training manual may generate an inaccurate explanation; the same image technology that supports design may also be misused to misrepresent events.

AI is not automatically transformative in every setting. Value depends on the task, quality of implementation, user trust, integration with existing processes, and whether the result is better than a simpler alternative. A spreadsheet, checklist, database query, or well-designed process may be more reliable and economical than an AI model for a narrowly defined task.

1.4 A useful mental model

Think of AI as a capable but fallible assistant. It can suggest, summarise, classify, translate, draft, and help analyse. It does not inherently understand the world in the same way a person does, nor does fluent language guarantee truth. The user must supply context, check important claims, and remain responsible for decisions.

Good AI literacy means knowing what the tool is suited to, recognising uncertainty, protecting information, and understanding when to involve a qualified person. It also means noticing who benefits, who may be excluded, and what happens when the system is wrong.

1.5 Questions to ask about any AI system

Before using a tool, ask: What task is it designed to perform? What inputs does it require? What information does it retain? Can its output be checked? What happens when it is uncertain? Is a human decision-maker involved? Is the tool accessible to people with different needs? Does the organisation permit its use? What would be the consequences of an error?