Artificial intelligence (AI) is the field of building systems that carry out tasks we usually associate with human judgment: recognizing a pattern, answering a question, choosing an action, or producing language.
What counts as AI
An AI system takes an input, applies a learned or programmed method, and returns a useful output. The method can be a set of rules, a statistical model, or a neural network. The label “AI” describes the job, not one single algorithm.
| Kind | What it does | Example |
|---|---|---|
| Narrow AI | One well-defined job | A spam filter, a credit-risk score, speech-to-text |
| Generative AI | Creates new content | A written summary, a piece of code, an image |
| General AI | Flexible human-level skill across many jobs | A research goal, not a product you can buy today |
How today’s systems are built
- Rules. Experts write if-then logic. Useful when the decision is stable and fully known.
- Machine learning. The system learns patterns from examples instead of a hand-written rule for every case.
- Generative models. A model trained on huge amounts of text, code, or images produces new content from a prompt.
A practical example
A card issuer watches each transaction. Amount, merchant, country, and time of day are the inputs. The system flags a purchase that does not match the cardholder’s usual pattern and sends it for review. That is narrow AI: one decision, learned from past fraud and genuine spend.
What to remember
- AI is the broad goal: machines doing work that needs judgment.
- Machine learning is the main way modern AI is trained.
- Generative AI is the branch that writes, draws, or codes something new.
