From Noticing to Doing
You are walking down the hallway at school and you notice a puddle of water on the floor. Your brain gets that information from your eyes. And then — almost instantly — you step around the puddle. You did not have to think very hard. You noticed something, and your brain decided what to do. AI agents do the same thing. They notice something with their sensors, and then they decide what to do. Today we will find out how an agent gets from noticing to doing.
Noticing Is Only the First Step
Sensing the world is important, but it is only the beginning. After an agent notices something, it has to figure out what that information means. And then it has to choose what to do about it. Here is a story. A robot is rolling down a hallway. Its camera sensor notices a big red object straight ahead. The agent gets the color numbers from the camera. But just knowing there is a red object does not tell the agent what to do. The agent needs to ask: What is that red object? Is it a wall? A stop sign? A ball? Once the agent figures out what it is, it can decide: should I stop? Should I go around it? Should I pick it up?
Sensing tells the agent WHAT is happening. Deciding tells the agent WHAT TO DO about it. Both steps are needed. Sensing without deciding is useless — and deciding without sensing is dangerous!
Let us follow an agent through a whole notice-and-decide moment. A smart home assistant is listening. Its microphone sensor picks up sounds constantly. Most sounds it ignores — the hum of the refrigerator, a dog barking outside, the TV in the background. Then someone says: 'Hey, set a timer for five minutes.' The agent notices the sound pattern that matches its wake word. It decides: that is a command for me! It listens to the rest of the sentence. It understands the words 'timer' and 'five minutes.' It decides: I should start a five-minute countdown. And it acts — the timer starts. From hearing sound, to understanding the command, to starting the timer. That is the whole journey from noticing to doing.
Flashcards — click each card to reveal the answer
When the Same Thing Leads to Different Actions
Interesting fact: the same sensor information can lead to different actions depending on what the agent is built to do. Imagine two robots that both see a tennis ball with their cameras. Robot One is a dog toy robot. When it sees the ball, it grabs the ball and brings it back — because that is its job. Robot Two is a floor-cleaning robot. When it sees the ball, it steers around it carefully — because running into the ball would knock it somewhere and make a mess. Same sensor information. Different decisions. Different actions. The agent's decision depends not just on what it senses, but on what job it has and what it knows.
Fill in the missing words.
When you see a puddle, you decide to step around it. When you see a trampoline, you decide to jump on it. Same information — your eyes see water or a trampoline — but your decision depends on what you know and what you want to do. Agents work the same way!
A robot's camera sees a ball on the floor. What does the agent need to do BEFORE it can decide what action to take?
Two robots see the same ball. One grabs it and one goes around it. What explains the difference?
Notice and Decide Cards
- You will practice going from noticing to deciding.
- Make six cards. On each card, write one thing an agent might sense — for example: 'Camera sees a stop sign' or 'Microphone hears a loud crash.'
- Now pick a job for your pretend agent. For example: delivery robot, home assistant, or guard dog robot.
- For each card, decide: what should THIS agent do when it notices that thing? Write the action on the back of the card.
- Shuffle the cards and try it again with a DIFFERENT agent job. Notice how the same sensing leads to different actions!
- Talk about it: why does the agent's job matter so much for deciding what to do?