Like many people, I used to spend a long time scrolling through stock image libraries, looking for the perfect photo or graphic to represent my idea. I’ll admit it: I don’t anymore. I have mixed emotions about this.

Three years ago, my cousin put together a video showcasing how he made his first $10k selling stock photography, and he’s earned a lot more since. So I think of him and so many other artists every time I choose to generate the exact image I need, instead of spending time searching for someone else’s licensed work. Really.

Decisions: AI or Human Assistance

I haven’t abandoned human creators. Not even close! When I needed a website designer and book copy editor, I chose humans over AI. I trust AI to make images for me because I’m the human writing the prompts and deciding when to use the images or not. I don’t trust AI to copy edit my book, because AI still makes mistakes. And I worried there would be errors in the website code and I’d spend too much time fixing things if I used AI.

This got me thinking about how educators can model thoughtful AI decision-making. We’re not just making personal choices about convenience or cost—we’re teaching a generation of students how to act and what to do. Our choices matter beyond just our own work.

I had a great conversation about AI imagery with the team from Pixabay while I was at the creator’s event at Canva Create last April. Their thinking was that images will become much like other creations – a world in which artisan items created by humans will be uniquely priced, valued, and collected. Interesting.

Testing the possibilities

My favorite image generation tool is Midjourney. Much of the time, spending a few minutes writing a good prompt will give me an image that is beautiful, fitting, and immediately usable. If I need any tweaks, I tend to use Adobe Lightroom for some things and Canva’s image editor for others. (I think Adobe’s filters are better, while Canva’s Magic tools are outstanding.)

As I evaluate new tools like Nano Banana—Google’s new AI-powered image generation and editing in Gemini 2.5 Flash—I’m applying this same decision-making framework. People everywhere are raving about how powerful it is, so I tried it out to see where it fits in my workflow. (I use the free version for testing.)

Awesomeness: The ability to change components of an existing photograph, illustration, or graphic to meet your needs. For example, I had Midjourney create a photograph using this prompt: a teacher standing on the right side of the classroom working with elementary students on laptops, cheerful bright colors, photograph –ar 16:9

Then I uploaded the image to Gemini and – over the course of 4 iterations – asked it to change the background wall to orange, change the teacher to female, modify the color of the student hoodies, and “Add a poster on the orange wall that has words “Welcome to Second Grade” and a sunshine.” Swipe through the gallery to see the evolution.

This is incredibly helpful, because I often find myself wishing an AI-generated image had small changes but is otherwise spot on. In my experience, my (paid) version of ChatGPT does an okay job at using reference images and editing. Nano Banana definitely wins in this department.

Disappointments: I had some major fails in my testing as well. Gemini couldn’t render an existing image in a new resolution, something I expected. When I had it combine two images, the output was something I could have done in Adobe Photoshop or Canva in just a few clicks, and it failed at generating something truly unique.

This testing reinforced my thinking: AI tools aren’t automatically better—they’re just different options that require thoughtful evaluation. But the bigger questions remain: How do we help students think about whether a job is best done by a human or by AI? How do we encourage them to consider partnership rather than replacement?

Reflections as an educator

Before reaching for AI, I want students to ask: Is this a task where human expertise matters more than convenience? How are our choices affecting the small businesses and creators in your community? What decision-making process are you modeling for others?

I don’t want everyone to immediately go to AI to perform a task. I want them to think about how their choices impact real people—like my cousin, like local designers, like the artists whose work built the foundation that AI now learns from. But I also know that the technology we have available can be efficient, cost-effective, and useful.

So what guidelines are you developing for ourselves? And more importantly, what are we teaching the next generation about making thoughtful choices in an AI-powered world?