Patterns, Rules, and Mistakes

What You Will Learn

• How AI uses patterns to make decisions
• How simple rules and learned patterns are different
• Why AI sometimes makes strange or funny mistakes

Patterns Everywhere

A pattern is something that repeats in a way you can notice.
Humans are very good at spotting patterns: shapes, sounds, habits, and even moods.

AI also works by finding patterns.

• It sees patterns in images that match things like “cat,” “tree,” or “car”
• It sees patterns in words that match questions, answers, or translations
• It sees patterns in numbers that match buying habits, health data, or game moves

The more examples AI sees, the more patterns it can find.

Rules Versus Learned Patterns

There are two main ways computers make decisions.

• Rule‑based systems use rules that humans write
• Learning systems build their own internal patterns from data

A rule‑based system might say:

• If score is above 90, then “A”
• If the light is red, then “stop”
• If password matches exactly, then “allow login”

These are clear and simple.
But some problems are too complicated to describe with a small set of rules.

Instead, AI models learn patterns through training.
They do not get exact rules like “all cats have this ear shape.”
They learn fuzzy patterns like “these combinations of pixels usually mean cat.”

Why AI Can Be Wrong

Because AI learns from data instead of fixed rules, it can be confused by unusual examples.

• A handwritten number that looks messy
• A picture taken from a strange angle
• A sentence with sarcasm or jokes

AI might still make a guess, but the guess can be wrong.

Sometimes the mistakes are funny, like calling a dog a muffin because of similar shapes.
Sometimes they are serious, like misreading a medical image.

AI is not trying to be silly or mean.
It is simply following patterns that worked most of the time in its training data.

Hidden Patterns and Accidental Rules

AI can also learn patterns that humans did not intend.

• If all cat photos in training data are indoors, it might secretly connect “indoor lighting” with “cat”
• If many dog photos include grass, it might connect “green background” with “dog”

In these cases, the AI is not learning “catness” or “dogness” the way humans understand it.
It is picking up accidental rules from the background and style of the images.

These hidden patterns can cause big mistakes when the AI is used in real life, where backgrounds, lighting, and styles change a lot.

When Patterns Are Not Enough

Patterns are powerful, but they are not everything.
Some decisions need reasoning, values, and context.

For example:

• A pattern might say “this person looks like past people who repaid loans,” but fairness might say “do not judge only by past biased data”
• A pattern might say “most people who click this also like that,” but safety might say “do not recommend harmful content”

Patterns alone do not know what is right.
They also do not know when to say “I am not sure.”
That is why human judgment and clear rules are still needed, especially in serious situations.

Summary Points

• AI uses patterns in data to make decisions and predictions
• Rule‑based systems follow human‑written rules; learning systems build their own patterns
• AI can make mistakes when examples are unusual or different from training data
• It can also learn accidental patterns that humans did not mean to teach
• Patterns are powerful, but they do not replace human reasoning, values, or responsibility

Practical Tasks

• Think of a time when you or someone you know misread a situation because of a pattern in your mind and write a short explanation
• Write a simple example of a clear rule a computer could follow, and an example of a task where patterns would work better than simple rules