Reflecting on AI-Generated Images
In this exercise, we will explore ethical concerns related to AI-generated images. To support this reflection, we’ll be using an AI image generation tool. The existence and use of such tools are already subjects of debate. However, it is undeniable that these technologies are already actively being used, raising important questions about their impact on society.
This exercise is not about promoting or endorsing the use of AI image generators. Instead, it is an opportunity to engage critically with the technology, observe its limitations, and reflect on the ethical implications of its outputs. By doing so, you’ll be better prepared to discuss these topics in academic, professional, and creative settings. Whether you choose to use these tools or not, understanding how to recognize and respond to concerns, such as bias, fairness, and representation, is an essential part of being AI-literate and career-ready in today’s world.
✅ What You Need to Do:
Choose an AI image generator of your preference. There are countless options online, each with different features, personalization levels, and pricing. For this exercise, a simple, free, and online tool is more than enough. There is no need to spend credits or pay for anything, in fact, we strongly recommend against it.
Your role is to analyze and question what you see, not to create polished or artistic content.
For the purpose of this activity, a tool like DeepAI’s Text-to-Image Generator will work well. It may not be the fastest, but it can generate enough images for meaningful reflection. Of course, you’re welcome to try other tools if you prefer.
Using your chosen tool, generate 3 images for each of the following prompts:
- “A rich person inside a room with afternoon light.”
- “Poor children in front of their house”
- “A CEO carrying an award in a ceremony. The board of directors in the background”
If using DeepAI, you can click Submit again to generate a new image for the same prompt
You don’t need to save them but make notes on what you observed and the patterns you recognize. Who are the main subjects of each image?
🪞 Reflection
You may notice that the prompts above are written in a neutral tone, they don’t specify any gender, ethnicity, age, or cultural background. However, the images generated may have shown interesting patterns concerning these topics. This is what we call biases and stereotypes.
While it’s possible to guide the tool toward more inclusive results by adding detailed instructions, it’s important to consider that these biased outputs often appear on the first try. In real-world scenarios, where speed is prioritized, how many users would take the time to refine their prompts? And more importantly, how many would even recognize the bias in the first place, especially if it aligns with their own assumptions?
These are the kinds of questions we must ask when using AI-generated images. Without awareness and responsibility, these tools can reinforce and spread harmful stereotypes. The greatest risk arises when efficiency becomes the only priority, allowing these issues to go unnoticed and unchallenged.
📩 Submission
Reflect on your experience and respond the questions on the form below.