What is AI?

What does AI mean?

AI means artificial intelligence. "Artificial" means made by people. The term was coined in 1955 by John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon, in their proposal for a research conference at Dartmouth College in the United States the following year.

The EU AI Act defines an AI system as a machine-based system that infers from the input it receives how to generate outputs such as predictions, content, recommendations or decisions. The core idea is that the system does not just follow a fixed instruction but works out the answer itself.

How does AI work?

AI works by learning patterns from examples. An AI system that should recognise cats sees hundreds of thousands of images labelled "cat" and "not cat", and adjusts millions of numbers inside itself until it usually guesses right. That method is called machine learning, and the technique with many layers of such numbers is called neural networks or deep learning.

A language model like ChatGPT learned the same way, but from text. It has read a large part of the internet and works out which word is most likely to come next. The answer is built word by word. That is why it sounds human, and why it can be wrong in the same confident tone.

What types of AI are there?

AI is often described in layers, where each layer is part of the larger one. Artificial intelligence is the whole field. Machine learning is AI that learns from data. Deep learning is machine learning with large neural networks. Generative AI is deep learning that creates new content, and language models are generative AI for text.

All AI that exists today is narrow AI: it is good at what it was trained for. An AI that can do everything a person can, often called AGI (artificial general intelligence), does not exist yet. Researchers disagree on when, and whether, it will come.

Examples of AI in everyday life

Examples of AI in everyday life are in most things you do on a phone or computer:

• Chat services like ChatGPT, Claude, Gemini and Copilot, which answer questions and write text.
• Face unlock on your phone and sorting photos by person.
• Recommendations on Netflix, Spotify and YouTube.
• Translation in Google Translate and DeepL, and live captions in video calls.
• Spam filters in email and fraud checks at the bank.
• Map apps that work out the fastest route from current traffic.

What is the difference between AI and ChatGPT?

ChatGPT is an AI service, and AI is the whole field. ChatGPT is built by the American company OpenAI on a language model, a kind of generative AI. Claude from Anthropic, Gemini from Google and Copilot from Microsoft are other AI services of the same kind. Calling ChatGPT "AI" is correct, but not all AI is ChatGPT.

What are the risks of AI?

The biggest everyday risk with AI is that it sounds sure when it is wrong. A language model can make up facts, sources and quotes, which is called hallucination or confabulation. Check anything that matters.

Other risks are that what you type is stored by the service, that AI can be used to create fake images and voices, and that an AI system can treat people unfairly if it learned from skewed data. The EU AI Act therefore regulates AI by how large the risk is.

How do I learn AI?

You learn AI best by using it. Start with a free AI service such as ChatGPT, ask about something you already know and see where the answer is good and where it falls short. Agentakademin is a free AI course in Swedish that takes you from the first question to your own AI agents, with no prior knowledge and no account.

Common questions about AI