Lesson 63 · 2 min
AI that sees your photos
Many chatbots can look at a photo, describe what is in it and read the text it contains.
Snap a picture and ask a question about it. Modern models can name objects, describe a scene, read handwriting or printed text, and explain charts. Reading text from an image is called OCR.
Everyday uses are handy: translate a menu, total up a receipt, identify a plant, read an appliance's error code, or turn a photo of a whiteboard into notes.
It can still be wrong with confidence. Blurry shots, small print, unusual handwriting and look-alike things, like edible and poisonous mushrooms, cause mistakes. Never rely on a photo answer for medicine doses or anything that could hurt you.
Photos also carry private details. Before you upload, crop out faces, ID cards, addresses and screens with personal data, and think about whose picture it is.
Do it now
Read a receipt
- 1. Take a clear photo of a shopping receipt.
- 2. Upload it to a chatbot with the prompt below.
- 3. Add up two or three lines yourself and compare.
Read this receipt. List each item with its price in a table, then give the total. Mark any line you could not read clearly instead of guessing.
Quick check
Keep going
Next →How AI shows it is unsureWords from this lesson
- Computer vision
- AI that makes sense of images and video, such as reading signs, spotting faces or checking products on a factory line.
- OCR
- Turning text in a photo or scanned page into editable, searchable text.
- Multimodal
- Able to work with more than one kind of input or output, such as text, images, sound and video together.