Maya watches her son type a math question into the homework helper. It answers with a neat set of steps.
"How does it know this?" he asks.
Maya is not sure what to say. Does it know math the way his teacher does? Or is something else going on?
AI4K12 is a project of two education groups, AAAI and CSTA, that sets out five big ideas for teaching AI.
One of those ideas is that computers can learn from data. Machine learning is a kind of statistical inference that finds patterns in data.
To find those patterns, a system needs a very large amount of data. This training data must usually be supplied by people.
| Step | What happens |
|---|---|
| 1. Collect | People gather a very large set of examples: text, photos or recordings. |
| 2. Train | The system looks for patterns across all those examples. |
| 3. Use | Given something new, the system uses the patterns to predict or suggest. |
AI4K12 adds an important point. AI systems can reason about very complex problems, but they do not think the way a human does.
So the homework helper is not a tutor who understands your child. It is a system that found patterns in a huge amount of text.
That is why its answers can be useful and still be wrong. You will see much more of this in week 2.
| Statement | True or false? |
|---|---|
| Machine learning finds patterns in data. | ? |
| Training data is usually supplied by people. | ? |
| An AI tool thinks about a problem the same way a person does. | ? |
| A machine learning system needs only a few examples to work well. | ? |
Sources: AI4K12, Five Big Ideas in Artificial Intelligence (poster, version 2)
Tomorrow: generative AI, the kind that writes and makes pictures, and how it chooses each word.