Generative AI: what it is and how to use it safely

What generative AI is

Generative artificial intelligence is the kind of AI that creates new content from a written request: it answers questions, drafts emails, summarizes documents, translates, and generates images and even video. ChatGPT, Gemini, Claude and Copilot are well-known examples.

These systems learn patterns from huge amounts of text and images. When you ask something, the model works out which sequence of words is most likely to make a good answer. It does not "look up" an encyclopedia for each sentence, which explains both how fluent it sounds and why it gets things wrong.

Read also: Deepfakes: how to spot fake videos and voice clones and AI in WhatsApp: what changes in your everyday conversations.

Why AI sometimes makes things up

Researchers call it hallucination when a system produces information that sounds convincing but is false or unsupported. A review in ACM Computing Surveys pulled together studies on the problem and showed it appears across many tasks, from summaries to question answering.

The U.S. National Institute of Standards and Technology, NIST, lists this risk, which it calls confabulation, among the main risks of generative AI, alongside leaks of personal data and misleading content produced at scale.

Source: Ji et al., Survey of Hallucination in Natural Language Generation, ACM Computing Surveys, 2023; NIST AI 600-1, 2024.

What it is good for in everyday life

• Turning scattered notes into an organized text.
• Summarizing a long document so you can decide what to read carefully.
• Explaining a hard topic in plain language, with examples.
• Drafting lists, travel itineraries and weekly meal plans for you to adjust.
• Fixing spelling and making a message clearer or more polite.

In all of these, you stay the editor: the AI suggests, you check and decide.

Safety rules

• Do not paste sensitive data. Skip passwords, card numbers, Social Security numbers, medical results and client information.
• Check facts, numbers and quotes against official sources before using them, especially for health, money and legal questions.
• Ask for sources and open the links. If the link does not exist or does not say that, drop the information.
• Review the privacy settings, which usually let you turn off the use of your chats for training.
• Be wary of apps promising "free unlimited AI" that ask for full access to your phone.

See also how to spot deepfakes and what changes with AI in WhatsApp.

How to ask better

The more context, the better the answer. Say who you are, who the text is for, how long it should be and the tone you want. Instead of "write an email," try "write a short, polite email to my landlord asking to fix the heating, which has been broken for a week." If the answer misses, ask for changes in the same conversation.