The Changing Nature of Information and How That Affects Instructional Design

AIs can generate words but they can't evaluate the truth of their statements. AIs can generate images and videos, but they can't evaluate the truth of what they generate. Sure, they can be asked to fact-check themselves. If you get an agent involved to do that automatically, you don't even have to ask them. But they can't tell a truth from a lie. AIs don't have the capacity to understand the difference.

Even if they could, they aren't often being used to fact-check or generate real information, they are instead often being used to generate things that feel right rather than are right.

Photo editing has been around almost since the invention of photography (see https://faculty.cc.gatech.edu/~beki/cs4001/history.pdf for examples). Manipulating images to evoke reactions is nothing new. Writing text to provoke reactions instead of to inform is a large part of the point of fiction.

What's different now is the speed and ease with which misinformation can be made. Sometimes it's about current events (https://www.yahoo.com/news/world/articles/ai-generated-video-misleadingly-shared-101020307.html), sometimes about health information (https://www.theguardian.com/technology/2026/jul/27/misleading-ai-generated-doctors-public-safety-danger-tiktok) and it's presented as fact. This happened before AI, too, but it's going to happen more and more often as it becomes easier and faster to make real-looking videos and real-sounding audio.

Because of the amount of AI-generated content online (and in print, too! https://www.atlasobscura.com/articles/ai-foraging-guides), and AI’s inability to understand what the truth is, we should raise our standards for what information we believe. Verify first, trust later.

As far as I'm concerned, my job deals in truth. If I add misinformation into a job aid for heavy machinery because an AI told me to, I am responsible when someone gets hurt after following it.

I'm not opposed to AI. I use it in moderation at my job. When I use it for research about any topic, I have it list the sources it used and I read them myself to make sure they actually contain the information the AI claims they do (spoiler alert: they often don't). I think it's a useful tool that is in the process of improving. I also think that if we lean on it instead of the good judgement of experts and ourselves, people are going to die. As a matter of fact, some already have: https://en.wikipedia.org/wiki/Deaths_linked_to_chatbots