AI Domains and Applications · Principle 6 of 6

Which part of AI is it?

Telling vision, language and data science apart.

Think

One AI or many?

A phone can unlock with your face, write what you say, and guess how long a bus will take. Does it use the same kind of AI for all three?

AI domains are sorted by the main kind of data they use: pictures, words or numbers.

A phone can unlock with your face, write what you say, and guess how long a bus will take. Does it use the same kind of AI for all three?

AI has three main branches, called domains. Each card is a job. Tap the group for the domain that does it.

Ask yourself: what goes in, pictures, words or numbers? Tap a group. Read the reason, then press Next card.

You can often tell the domain by asking what goes in: pictures, words or numbers.

Some apps use more than one. The signboard reader first looks at a picture, then works with words.

AI domains are sorted by the main kind of data they use: pictures for vision, words for language, numbers for data science.

The three are computer vision, natural language processing and data science.

AI domains are sorted by the main kind of data they use: pictures for vision, words for language, numbers for data science.

Cameras and microphones collect personal data. Ask a trusted adult before you allow them.

A railway office studies a table of last year's delays at each station to plan next year's timetable. Which domain is this?

A photo app lets you type the word “beach” to find your beach photos. Which domains does it use?

A game app asks to use your phone's camera and microphone so that its AI can work. What should you do first?

Computer vision, because railways are full of things to see. No pictures are used here. The data are tables of numbers, so vision is not the main part.
Language, because station names and notes are written down. The table holds numbers and names, not sentences to read, so language is not the main part.
Data science, because the main data are numbers in a table — correct. Yes. Looking for a pattern in a table of numbers and using it to plan is data science.
Only data science, because an app is full of numbers inside. Apps are made of many parts. Here, one part reads your word and another looks at the photos.
Language to read the word, and vision to look at the photos — correct. Yes. Many real apps mix domains: one part works with words and another with pictures.
Only language, because the word you typed is what matters. The typed word is only the input. The app must also look at the photos, and that is vision.
Tap yes, because a game app cannot work without them. Many apps ask for more than they need. A camera collects pictures and a microphone collects voices, so check why first.
Tap yes, because all my friends already allowed it for this app. Your friends may not have thought about it. Each of us should check before sharing pictures or voices.
Ask a parent or teacher, and check why the app needs them — correct. Yes. A camera gives pictures and a microphone gives voices, so a trusted adult can help you decide.