Wait...That's not right... - Beth Wolf Blog Post #9

 One of the quotes from the readings that stuck out to me this week was how people don't think about a post's accuracy until they are prompted to do so. Once they are prompted, they are right a majority of the time. But that simple act of the ask is what is missing in most people's every day social media scrolling.

This topic comes up every so often in my friend group chat, especially now that AI is so prevalent. Not about the accuracy of certain news articles or whether Trump did this certain thing. They can usually spot those stories and they are very smart people when it comes to spotting a reliable versus non-reliable news source. But when we share videos with each other from TikTok, sometimes we get into a debate over whether it is real or not, and it is all because of AI.

If you do a Google search for AI video, the top result wants you to search for AI video generator. All of the website results are AI video generators, even though I didn't specify that. These AI videos depict animals, humans, full Hollywood style feature films, whatever you can dream, they can generate. But they always seem slightly off.

Don't get me wrong, some of them are very good and you almost can't tell that they are AI generated. But that almost makes me mad that I was duped into watching this video thinking it was real and then I find out that it was never real. So I had to start to teach myself how to spot the inconsistencies right away so I didn't get into that situation anymore.

Here are some tips I gleaned from a few different sources:

1. Pause the video when there is movement, especially fast movement. If the movement looks unnatural, blurred, or completely wrong when it is paused, it's a good chance that the AI didn't bother filling in the gaps between stopping points to make the moving object look natural in each frame.

2. Lighting and shadows that look unnatural or out of place. AI doesn't always think about the shadows that objects will cast in the room so this is a dead giveaway.

3. Unexpected background noises, such as nature sounds when the video takes place in a city. Or no background noise at all.

4. Body language and facial expressions. Human beings are so complex and very hard to recreate without some mistakes. Pay attention to unnatural facial expressions, unnatural movements, fingers that don't do what fingers should do, and even a flat sounding voice with no emotion, etc.

AI videos are getting better and better every day. Technological advances in AI are making it almost impossible to use these tips and I'm sure in a couple months, this post will be obsolete. The thing that scares me the most is that people will use AI to do real harm with misinformation, disinformation, or malinformation. So whatever you see on a video, regardless of if it is an AI video or not, make sure you verify that the content is real with multiple sources or with tips from the Verification Handbook reading from this week!

Question for the reader: Have you been fooled by an AI video lately? How did you find out?

Comments

Popular posts from this blog

Social Media for Libraries: Training for Library Marketing

Libraries Reaching Rural Communities Through Social Media

A Missed Opportunity? - How libraries could use YouTube to increase visibility - Beth Wolf Blog Post #1