Free · Runs in your browser · No account

The AI Lab

Three demos that open the hood. Everything runs on your own device: nothing you click is sent anywhere, and no sign-up is needed. Safe for every age.

Teach the machine Predict the next word Expert system

Demo 1 · Pattern recognition & learning

Teach the machine

Pick a color, then click the board to add examples, like showing a computer pictures labeled “apple” or “orange.” The machine looks at the closest examples it has seen and guesses. That is real machine learning (it is called k-nearest neighbors).

Neighbors

Try this: load the example, then notice the one gold dot sitting in cyan territory. Switch neighbors from 1 to 5. With 1, the machine “memorizes” that odd example; with 5 it ignores it. That trade-off is a big part of real AI engineering, and a lesson about bad training data.

Demo 2 · Natural language processing

Predict the next word

This tiny language model has read only Psalm 23, Psalm 8:3–6 and Psalm 139:14. It counts which word tends to follow which. Chat assistants work on the same idea, just trained on vastly more text with far bigger math: they predict likely next words.

Start with:

Choose the next word:

Notice: press “Let the machine write” a few times. It makes sentences that sound like the Bible but are not in it. It is not lying on purpose; it does not know what is true. It only knows what usually comes next. That is why we always check the source.

Demo 3 · Expert system (requirements 9–11)

Which honor should I earn next?

An expert system does not learn from data. A person writes the rules, and the computer applies them and explains why. Answer the questions and watch the rules.

Do you love being outdoors?

Do you want to fly something?

How old are you?

Do you like building things?

Some questions do not matter for some answers. Real expert systems only ask what they need. Notice how this one shows its reasoning: that is the big advantage over a “black box.”

For instructors

What each demo teaches

Demo	Honor requirements	The lesson
Teach the machine	5 (learning, pattern recognition), 7	AI learns from examples, so bad or unbalanced examples produce bad guesses.
Predict the next word	5 (natural language processing), 7, 8	Language models predict likely words. Sounding right is not the same as being true.
Expert system	9, 10, 11 (knowledge engineering)	Rules written by an expert, applied consistently, with the reasoning shown.