A Look into how AI is Advancing
- Logan Coyle
- 1 day ago
- 2 min read
Reporter: Coyle Logan
PHOENIX — As generative artificial intelligence becomes more realistic, automated images and videos are increasingly filling social media feeds.
For Andrew Dietrich, an Arizona State University student, the saturation of AI content in his feed eroded his trust in social media. In response, he chose to delete his accounts.
AI-generated content is becoming increasingly difficult to detect. Even as production surges, studies show users can barely tell what’s real, something ASU students say they experience daily.
“It's just becoming less and less human-based and more and more automated,” Dietrich said. “We're losing the human aspect of it, which is what made it so good in the first place.”
Dietrich isn’t the only one who’s realized the influx of AI-driven content on his feed.
An Artsmart analysis estimates that about 71% of images shared on social media worldwide may be AI-generated.
Before Dietrich quit social media, he said that “1 in 5” videos on his feed were AI.
He was confident in his ability to detect generated videos, saying that watermarks and other small signs are telltale giveaways.
However, a study by Runway found that people often overestimate their AI detection skills. Participants deciphered between next-generation AI and real videos at 57% accuracy, a rate slightly higher than chance.
When Dietrich took the Runway test, he scored a 50% detection rate. His answers were “mostly guessing,” he said.
Runway’s results are convincing, but what separates its test from what is present on social media is the lack of a comment section.
“The real reason I believe a bunch of videos are AI is because I think so, but then I go to the comment section, those comments also say that's AI,” Alexander Jaus, ASU student, said.
Without a comment section, Jaus, who scored a 5/20 in the Runway test, said it was much more difficult to detect realistic automated videos.
He said he typically sees multiple AI-generated videos in his feed on any given day, but they are not as realistic as what he saw on the test.
“I see videos all the time, these people melting into the earth and doing these extravagant things, and I just don't think how people can find that real,” Jaus said.
The amount of content being produced continues to surge; a Forbes report estimates that about 34 million AI-generated images are created each day.
Sean Mowery, an ASU student, likes to generate pictures and videos of his friends for fun, he said.
The AI content that appears on Mowery’s algorithm is often absurd, he said, and he could easily detect it.
However, he expressed concerns that older generations have a harder time distinguishing between real and AI video.
“My parents, they'll send me something that's like obviously AI, and they can't tell, and I have to tell them that it's AI, and sometimes they don't even believe me,” Mowery said.
For some students, like Dietrich, the inflow of automated content signals something deeper than convenience.
“I search for authenticity in what I do and what I interact with, so anything AI-generated to me just seems like a lack of effort,” Dietrich said.


Comments