Week 11 Ā· Unit 9.2: My childrenās digital world Ā· 3 lessons
2
Why AI took off
Digital Literacy 9: Unit 9.2, objectives 9DW.06, 9TC.01
GoalExplain why AI became useful so quickly in recent years, using data, computing power and better methods.
New words
data Ā· facts, numbers, pictures, sounds and text that can be stored and usedprocessor (chip) Ā· the part of a device that carries out calculationscloud computing Ā· renting computing power and storage from big computers reached over the internetalgorithm Ā· a clear set of steps for solving a problem
Explanation
People have dreamed of thinking machines for a very long time, and early AI programs were quite limited. AI became far more useful recently because three things came together. First, phones, cameras, sensors and websites produce huge amounts of data to learn from. Second, chips became faster and cheaper, and many of them can share one big job. Third, better learning methods made use of the data and the chips. Cloud computing lets even a small team rent that power without buying machines. Think of baking: you need a recipe, ingredients and an oven, and a missing one spoils the cake. There are costs too. Training uses a lot of electricity, and the data may hold personal details about real people.
Worked examples
A pronunciation checker makes fewer mistakes as it learns from more recordings. This chart is invented, but the pattern is typical: more examples usually help, though it never reaches zero.
Data
Chips
Method
few photos
plenty (rented)
good
A school robotics club wants a camera to sort recycling. It has a good method and a cloud account, but only a few photos. The idea will work poorly until it has more examples.
Cloud computing lets a small team share power. Many laptops send their jobs to big computers and get the answers back.
Class activity
The ingredient game (groups of three). Each person holds one card: data, computing power or learning method. Your group is given a goal, such as āsort fruit by photoā or ātell if a cough sounds unusualā. Each person says in one sentence what their card must provide for this goal. Then the teacher removes one card at random: what goes wrong without it? Finish with a short class debate: if companies need lots of data to build AI, who should be allowed to collect it, and what should they promise in return? Give a reason for each opinion.
Practice
1
A village team has a strong learning method and fast rented chips, but only 40 example photos. Which ingredient limits them, and what would you advise?
Team has
Result
method, chips, 40 photos
weak guesses
Data is the limit. They need many more varied photos, collected with permission from the people or places shown.
2
True or false, and fix if false: āNobody had thought of AI before recent years.ā
False. People imagined thinking machines long ago. What changed recently is that data, chips and methods made it practical.
3
Why does cloud computing help a small school team that wants to try machine learning?
They do not need to buy expensive machines. They rent power when they need it and pay only for that, so small groups can try ideas.
4
Name one worry about the data used to train AI and one way to reduce it.
Worry: the data can contain personal details, photos or voices of people who never agreed. Protections: ask permission, remove names and faces, collect only what is needed, and follow data protection rules.