What an Agent Notices
Imagine you walk into the kitchen and notice: the stove is on, something smells like burnt toast, and the smoke alarm is beeping. In about one second, your senses collected all that information and your brain put it together into a picture of what is happening. AI agents do something similar. Their sensors collect information from the world, and the agent puts it together into a picture it can understand. Today we will find out how that works.
From the Real World to Numbers
Here is something important to understand. Computers do not see the world the way you do. They do not see colors. They do not hear music. Everything a computer works with is numbers. So when a camera takes a photo, the AI agent does not see a picture like you do. Instead, it sees a huge grid of numbers. Each tiny dot in the photo, called a pixel, has a number for how red it is, a number for how green it is, and a number for how blue it is. A photo that looks beautiful to you is just millions of numbers to the agent. The same is true for sound. When a microphone picks up your voice, the agent gets a long list of numbers that describe how the air was vibrating at every tiny moment. It does not hear music — it sees patterns in numbers.
Sensors turn real-world things — light, sound, temperature, touch — into numbers that the agent can work with. Everything an agent notices is really just information written in numbers.
Let us follow a single piece of information on its journey into an agent. Step 1: Something happens in the world. A dog walks in front of a robot's camera. Step 2: The camera sensor captures the scene. It turns the image into millions of numbers. Step 3: The numbers travel into the agent's brain — its computer processor. Step 4: The agent looks at those numbers and tries to understand them. Are these numbers the pattern for a dog? Or a cat? Or a tree? Step 5: The agent decides: that is a dog! All of that happens in a fraction of a second. The sensor did step 2. The agent did steps 3, 4, and 5.
Fill in the missing word to complete each sentence.
Noticing Many Things at Once
Some AI agents have just one sensor. A simple thermostat only has a temperature sensor. It only needs to know one thing: how warm is it right now? But many agents have several sensors working at the same time. A self-driving car uses cameras to see the road, microphones to hear horns and sirens, GPS to know its location, and special sensors called LIDAR that shoot out laser beams to measure exactly how far away other objects are. All of those sensors are sending information to the car's agent at the same moment. The agent has to put all that information together — like fitting puzzle pieces — to understand what is happening around the car right now.
Match each piece of real-world information to what the sensor turns it into for the agent.
Terms
Definitions
Drag terms onto their definitions, or click a term then click a definition to match.
The more sensors an agent has, the richer its picture of the world. A robot with only one camera might not notice a banana peel on the floor. But a robot with cameras, touch sensors, and floor sensors would notice it from several directions at once!
How does a camera sensor show an AI agent what a photo looks like?
A self-driving car uses cameras, GPS, and LIDAR sensors all at once. Why does it need so many?
Be a Human Sensor
- Sit quietly for two minutes with your eyes closed.
- Listen carefully. What do you hear? Make a list of every sound.
- Feel the air on your skin. Is it warm or cool? Is there a breeze?
- When two minutes are up, open your eyes.
- Now write down everything you noticed with each sense: hearing, touch, and sight.
- Count how many pieces of information your senses collected in just two minutes.
- Talk about it: if you had to turn all of that into numbers for a computer, what would be hard? What would be easy?