Evaluate how primary sources influence the narrative of a historical or contemporary issue, and how different narratives can affect our understanding of the issue
- Investigate three instances of AI-generated images online: one that is harmless fun, one that is dangerous disinformation, and one that is somewhere in the middle. For example, fake viral images of explosions at the Pentagon and the White House, fake images of Trump’s arrest, or a fake image of LeBron James). The News Literacy Project keeps a list here. Write 2-3 paragraphs describing the instances of AI-generated images that you selected and their impact. How were they discovered to be fake, and how long did it take? What qualities make one instance harmless and another dangerous? What is the threshold between mis- and disinformation in these instances?
- Next, try to recreate the three images you selected using a text-to-image AI tool such as Adobe Firefly, DALL-E, Midjourney, or Stable Diffusion. See if you can create a more realistic version of each by improving your prompts at least three times each. For instance, you might provide a text-to-image AI tool with the following three prompts: “An image of [insert celebrity’s name] being arrested,” “A realistic photograph of [insert celebrity’s name] being arrested by four police officers on a street corner,” and “A hyper-realistic photograph of [insert celebrity’s name], wearing a dark blue business suit, standing in front of a courthouse with his/her hands behind his/her back as he/she is being handcuffed by police officers in black uniforms, crowds gathered around them, some people taking photos of the chaotic scene with their cellphones, natural lighting, low contrast.”
- Write a reflection including which tool you used, the prompts, and the resulting images. What was easy or difficult about generating the images? Were you successful in creating more realistic versions of the images? What artifacts give away that the images are AI-generated?