Technology & AI

A large language model learns statistical patterns from huge collections of text (and, for image or video models, pictures and clips). It learns which words tend to follow others in which contexts.

Text is broken into small pieces called tokens, often parts of words. The model reads your prompt plus the conversation so far.

The model picks a likely next token, adds it, and repeats until the answer is complete. It does not look facts up unless the tool is connected to search or documents.

Because the output is a prediction of plausible text, it can invent facts, figures or sources ("hallucinations"). Always verify anything important, especially money, health and legal matters.

Assistants, writing tools, image and video generators. See the AI tools hub, AI helper vs virtual assistant and AI video tools.

Use long, unique passwords for every account and a password manager to remember them. Try the password strength visual in the cybersecurity hub.

Turn on MFA, especially for email, banking and social media, so a stolen password alone is not enough.

Be suspicious of urgent messages asking you to click, pay or log in. Go to the site directly instead. See secure browsing.

Turn on automatic updates for devices and apps; updates fix security holes attackers use. See mobile security.