1. AI isn’t only a text box

Some assistants can interpret pictures, describe diagrams, accept spoken questions, generate images or transcribe audio. This is called multimodal AI: working across more than one type of input or output. Features, accuracy and limits vary.

2. Ask about an image with context

A photo of a kitchen cupboard could help you brainstorm storage ideas. A photo of a chart might help you describe its axes — but the assistant can miss small labels, misread a scale or invent details. For important measurements or documents, check the original yourself.

3. Voice makes interaction natural, not infallible

Speaking can make AI easier to use while walking or learning. Voice transcription can mishear names, dates and numbers. If you’re on the road, stay focused on driving and follow road-safety laws; never operate controls unsafely.

4. Generating images responsibly

For a fictional poster or concept sketch, describe subject, scene, style, colours and aspect ratio. Be aware that generated imagery can misrepresent real people or events. Check likeness rights, permissions and platform rules before public use, especially for marketing.

5. Be careful with recordings

A meeting or phone call may involve consent, confidentiality and company policy. Don’t record or upload other people’s voices without the necessary rights, permissions and safeguards.

See it in action

This example shows how to apply the lesson. You can reuse the structure with safe, fictional or public details.

Your practical exercise

Use a fictional image brief or a picture you own without private details. Ask the AI to describe what it can verify visually and what remains uncertain; then rewrite the result as an image brief. If your service lacks image tools, practise with text descriptions.

Check your understanding

Answer in your own words before opening each explanation.