Real or fake? What you need to know about the world of AI

CLEVELAND — Generative AI. Deep fake videos. Anyone who spends time online has, without a doubt, seen them. As many as eight to ten times a day.

Artificial intelligence is a technology that constantly learns and, as such, is getting better and better at being able to create images and videos that seem real, even if you can tell they are not.

But it is getting harder and harder to tell.

“I think the biggest difference between what’s happening now and what happened over the past few decades is that generative A.I. is evolving at a very fast pace, and we’re starting to see things now that the technology wasn’t able to do in just a couple of years ago,” said Daniel Alkenquer, Directof of the School of Digital Communication, Design, and Emerging Technology at Kent State University.

‘So utterly sorry’: Man who snatched ball from young girl at Guards game apologizes

For the most part, the images and videos that creators are posting to social media, even if they are not real, are meant to entertain.

Realistic videos that can use an individual’s real voice with a computer-generated image to have them say or do something that, at first glance, seems authentic.

Incredible videos in which you may see an animal doing something that you know is impossible or implausible.

Still photos in which details, including the expressions on people’s faces, can be convincingly altered.

“Computer graphics, it’s not really that new. I think we’ve all watched movies with visual effects that we couldn’t tell that if they were real or not,” said Alenquer.

Being able to tell if what we see is real or not is growing more and more difficult.

“I think human beings are really good at catching things right away,” said Alenquer.

“We can look at something, and sometimes it just doesn’t feel right. Right. To look at something and maybe it’s the motion, maybe it’s the sound of the voice, maybe it’s the rhythm. You know, there are these small clues, right, that sometimes we can’t catch them individually, but as a whole, we get this feeling that, hey, maybe this is not right,” he added.

In his classes, he shows students various photographs, some of which are and some of which are not completely AI-generated and asks students to try and tell the difference, then explain their conclusions.

“Technology is evolving, but I know from the students, and I have been doing this for three years, people are also getting better at recognizing AI images because there’s so much out there, and we get better at recognizing these small things and all of these details that are really important for us,” said Alenquer.

Whether or not it is inherently bad is not something to which there is a consensus.

“It’s interesting because the way that students reply there was almost no single answer for saying A.I. is good for ice. Evil, I think, was about 20% of the students said A.I. is evil. About 60 to 70% said A.I. is good and it’s evil in about 10%,”said Alenquer.

But being able to tell real from fake may be far more important than being able to tell if we have been fooled by a computer-generated image or video.

‘Deep-Fakes’ can replicate someone’s voice, merge it with a computer-generated image of that same person and have them seem to convincingly say or do something they didn’t.

In advance of November’s mid-term election, public office holders and political candidates are already watching for computer-generated videos of them meant to influence voters.

Manipulating viewers’ thoughts and actions.

Why gas prices rocketed up in Ohio overnight – what could come next?

And generative A.I. is also being used by criminals to lure vulnerable viewers into schemes.

Images of public figures and entertainers are altered to make them seem as though they are endorsing a product. When people attempt to purchase that product, they find out too late that the product doesn’t exist, and they have given their bank account information to the thieves who created the fake ads.

Scams and Schemes

“I use the Internet as our great gateway to all the information we ever wanted. But same is true on the bad actors front. They use the same internet we use to attack the same targets. And when we look at the prevalence, yes, it strikes us 8 to 10 8 to 10 times every single day,” said Chelsea Triboniak, a former U.S. Army officer who owns a cybersecurity company based in Lakewood called Critical Ops.

“The level of sophistication anymore in these scams is almost surreal. And that you have to study the objects, the image, the text to say is that fake and why? And at times you really cannot differentiate,” said Triboniak. “And anybody who says that they don’t fall for them, everybody does, for multiple reasons. We want sometimes that too-good-to-be-true scenario, to be too good to be true,” she told Fox 8 News.

To be able to see firsthand what generative A.I. can do and to try and teach people how to better tell real from fake, Triboniak had a ‘deep fake’ created of her, duplicating her voice and merging it with an entirely computer-generated replica of herself.

“All the hand postures, the voice inflection. I have a unique voice; things of that nature were very accurate. And after doing that, yes, you are alarmed at going, Wow, I know myself, and I can pick out the little nuances, but can somebody else,” said Triboniak. “And then it even showed me the voice they used, which was actually a young man. And I had to laugh because all of a sudden, you go, okay, well, that I can see a resemblance in the girl. The way her facial structure sits, the size of her forehead, things of that nature that are unique to you. But then to have a young man’s voice connected as mine, you kind of went, wait a second.

She believes it may be safest to question everything you see online, examine the source, and realize that if something seems too good to be true, it most likely is.

“Trust but verify”

Triboniak even advises people to remove their voice from the voicemail on their phone, explaining that, for criminals, there is enough of your voice there for them to replicate it and use it in a scheme.

“When we start to flirt with having no trust, which is kind of the cybersecurity side of this, versus human instinct of I want to be able to trust you, we reached this almost disconnect that makes it a struggle to society,” said Triboniak. “It’s just a tool within itself. It’s probably not good or not evil, but it’s about the intentions, right, of folks generating that content in the background,” said Alenquer.

“Nobody wants to be fooled, right? People want to know what they’re looking at. And if content is being generated to make people believe that it’s authentic, you know, people do care,” Alenquer said. “So the to be too good to be true is the first one that, frankly, all of us want to win the lottery. It’s not going to happen tomorrow. Unfortunately.”

“The second thing is really just to validate, when I look across to all the people in the image, have all five fingers or are they connected, are the ears finished? Which is difficult for generative AI to do,” said Triboniak. “At the end of the day, when you’ve been violated to that extent and the amount of pain and agony that goes into fighting the battle to get your identity back, even your money at that rate, can you get it back?”

When Cedar Point is opening next month: What’s new at the park?

Particularly if a transaction involves cryptocurrency, the answer is that it becomes very difficult or impossible to recover.

AI videos are entertaining. The images and scenarios are captivating, but they can also be criminal.

“If you’re spending your time (online) just slow down and second-guess, is this real or fake? Did I validate from multiple sources?” said Triboniak.

“It’s just a tool within itself. It’s probably not good or not evil, but it’s about the intentions, right, of folks generating that content in the background,” said Alenquer.

For the latest news, weather, sports, and streaming video, head to Fox 8 Cleveland WJW.