Artificial intelligence is getting tougher to spot in Kentucky: Media studies expert helps tell the difference

LEXINGTON, Ky. (FOX 56) — Artificial intelligence continues to evolve rapidly, especially in the media space. While it may have been easier to spot AI-generated material in recent years, it’s getting harder.

“The danger comes when we don’t believe anything, when we go from skepticism to cynicism. And that’s the real danger of AI,” Al Tompkins from the Poynter Institute for Media Studies told FOX 56 News.

It’s a technology that’s evolved a lot in a short amount of time and begs the question, can we always believe what is seen?

“All of those old, reliable detections. Like they’ve got too many fingers, or their eyes look weird or something, the depth of field in the photograph doesn’t make sense. All of those are starting to fall by the wayside,” Tompkins said. “Ask yourself, where did this come from? How did this start? Who’s behind it? And the second thing I would say to you is, remember that disinformation mostly turns on emotion.”

Tompkins has spent his career teaching journalists how to dig a little deeper and avoid being tricked when something appears real online. And there’s a lot online that can easily trigger a firestorm of comments and shares, sometimes by design.

 “We saw this, for example, with the California wildfires; we’ve certainly have seen it with the protests out in California, that sometimes we see images that align with what’s actually going on but the images might be from some other place, or they might not be real at all,” Tompkins said, pointing out it’s a lot easier to create deceptive material during breaking news situations or during conflicts like the Russia/Ukraine war and Israeli/Palestinian conflict.

Tompkins noted that when you have high emotions, judgment is running low. Some readers might remember an AI-generated image of a young girl that went viral during Hurricane Helene last October; she’s not real, but at first glance, it’s easy to assume she is. Late last month, Google unveiled an update to its generative video model, Veo 3. The technology can now create short, high-quality videos with natural sounds that are entirely artificial.

“I think it’s far more useful to educate people about how to use internet capacity, how to use artificial intelligence capacity. Education is the elixir for a lot of this. Not so much legislation,” Tompkins said, adding it’s especially important for less tech-savvy generations to keep up. He said that artificial intelligence detector technology has been mostly able to keep up, but especially with audio and video, it’s not quite foolproof.

And keep in mind there’s the extreme possibility that something that appears to be AI-generated at first actually is real.

“There are extraordinary things that occur that just don’t seem possible. You know, if we saw that video of the Hindenburg blowing up, I don’t think any of us would believe it until we had some eyewitnesses that were truly believable,” he said.