You and I + AI

Do

This slide includes quick links to today’s lesson materials. Teacher-facing slide. Use before class to open the lesson plan and Drawing Prompt Sheet. This slide can be skipped when presenting to students.

Say

Welcome to Expedition 3: Imagination and Innovation. In this expedition, we’re going to explore artificial intelligence, or AI, as another tool people can use to create. We’ll think about what AI can do, how it works, and how creating with AI is different from the way people imagine and create. Today we’re going to compare something humans do all the time—imagine and create new ideas—with the way AI can generate images from words.

Say

Here’s your challenge: Draw a shark with legs chasing a cat. You probably have never seen this exact scene in real life, so think about how you decide what to draw. What does a shark look like? What might shark legs look like? How will you show that it is chasing the cat. Take 90 seconds to draw your version. Now compare with someone near you. Did your drawings look exactly the same? What did you do differently?

Do

Give students a short amount of time to draw and maybe an additional 30 seconds. Follow this time by brief partner comparison. Ask 2–3 students to share differences they noticed. Use the discussion to surface the idea that people combine past experiences, visual memories, imagination, and individual choices when creating something they have never seen before.

Say

We’ll begin by using our own imaginations, then learn a little about AI and AI training. We’ll investigate how words, images, and patterns connect, and end by thinking about what AI can—and cannot—do. Lesson goals: Introduce AI and AI Training. Explain how AI can generate images off of a text prompt. Understand how LLMs are trained and how they generate images using patterns from data rather than thinking like a human.

Say

Now we’re going to experiment with the connections between words, images and patterns. Secretly choose one prompt from the list in front of you. Do not show your partner. Draw an image that communicates that prompt. Your challenge is to include enough visual information that someone else might figure out what your words were. When you’re finished, swap drawings with your partner. Study their drawing and write the prompt that you think best describes what they drew. Then reveal the original prompts and compare.

Do

Distribute the Drawing Prompt Sheet or prompt cards, paper, and pencils. Circulate and remind students that artistic skill is not the goal. The goal is communicating information through an image.

Say

Artificial intelligence, or AI, is computer programming designed to perform certain kinds of tasks. Some AI tools can recognize patterns. Others can generate things like text or images from patterns it has learned, when a person gives them instructions, called a prompt. You may have already used or seen AI tools. What are some things you have seen AI do?

Do

Take 2–3 quick responses. Keep the definition broad. Reinforce that AI refers to computer programs and tools rather than a single program.

Say

Now look at this dog. During training, AI systems analyze large amounts of data and learn patterns and relationships within that data. What are some visual features you notice? What other words could describe this image besides dog? Much of this learning happened by people making relationships first and labeling the image. Large language models were trained on data collected by humans labelling common features images like this.

Do

Ask students to generate several descriptors for the image such as fluffy, brown, small, furry, outside, or cute, smiling, collar, beagle like dog. Emphasize that a single image could connect to many words and descriptions.

Say

So how can an AI tool connect words with images? Example, imagine the system encounters many images that people describe with the word dog. Across those examples, there may be patterns involving things like ears, cute, or fluffy. The more relationships the system learns between words and visual patterns, the more information it can use when responding to a new prompt.

Say

What happened when you compared the original prompt with your partner’s guess? Were they always exactly the same? There are several reasons they might differ. People interpret words differently. The thing we imagine may not match what another person imagines. And when we draw, some information may be missing or unclear. Look at these two examples. If we switched the prompts (the yellow cards), could they still describe the drawings? Why or why not?

Do

Invite 2–3 responses. Ask students to point to specific visual evidence: What in the drawing makes you think shark, monster, cat, dog, or chasing?

Say

We’ve talked about patterns before. Think about one of these questions, get an answer in your head, then hold up that number share with someone else near you that is holding the same number. Finish with: How did you know something in your partner’s drawing was probably a dog or a monster? You recognized features you’ve seen before.A pattern is something that repeats or something we recognize happening again and again. AI tools also use patterns, including relationships between words and visual features. Does that mean AI thinks about a picture the same way you do? No. AI can recognize and use patterns, but that is different from having human experiences or imagination.

Do

Give students brief think time, then invite responses. Reinforce the distinction central theme humans bring experiences and imagination; AI systems generate outputs using learned relationships and patterns.

Say

Now consider two prompts: A monster chasing a dog or A monster and a dog. Almost all the words are the same. What changed? The word chasing adds information about the relationship and action between the monster and dog.

Do

Invite students to describe expected differences, such as: one character behind another running movement expressions distance or direction

Say

The word chasing adds information about the relationship and action between the monster and dog. This is why the details in a prompt matter. Changing a word can change which patterns and relationships an AI tool uses when generating an image. Additional descriptive words give the artist—or AI system—more information to work with.

Do

Connect back to students’ drawings earlier on slide 10

Say

Here is what happens when we search for images using the phrase a monster chasing a dog. Notice that these results are not all the same—and some don’t match the phrase especially well. What parts of the search phrase seem to show up most often? What seems to be missing or mixed up?

Do

Invite several quick observations from the image-search results. Clarify that an image search is finding existing images, not generating a new image. This prepares students to compare search with image generation on the next slide.

Say

Image generation works differently from image search. Instead of finding an existing image, an AI image-generation tool creates a new image in response to a prompt. Look at these two examples. One prompt says a monster chasing a dog. The second adds one word: happily. What changed between the two images? That one word gives the system another relationship to represent. It might change facial expressions, body language, colors, lighting, or the overall mood.This is another example of how changing a prompt can change the image an AI system generates.

Do

Ask students to compare the images using specific evidence rather than simply saying one looks happier.

Say

Early on, AI had a hard time telling the difference between blueberry muffins and Chihuahuas. We can see that there are some similarities in the patterns, color of the muffin or th the color of the fur, eye that look like blueberries. We call this technology artificial intelligence, but intelligence itself is a complicated idea. There is no single one way of being intelligent. Which one feels like the most important part of intelligence to you? There isn’t one correct answer.

Do

Have students show a finger vote from 1–5 or turn and talk. Invite 2–3 students with different choices to explain their reasoning. Use this as an opportunity to complicate—not settle—the question of what we mean when we call AI “intelligent.”

Say

Let’s pull together today’s big ideas.