Consumers can't tell the difference between human-made and AI-generated videos, study suggests

A survey of U.S. consumers indicated that a strong majority of Americans would be comfortable supporting government regulation that required labeling on artificial intelligence-generated content. The same consumers surveyed had difficulty discerning AI-generated video from human-made video content.

Americans responded to a HarrisX survey asking them if they wanted "U.S. Lawmakers to Require Labeling on AI-Generated Content," with most responding that they would support such an endeavor.

Consumers were asked about fully AI-created videos, photos, writings, music, captions, sounds, and more, Variety reported.

The strongest response came in regard to labeling AI videos and photos, at 74% and 72%, respectively. Only 61% of respondents supported labeling AI-generated sounds and captions; representing the lowest amount of support on the survey.

Even though the majority of consumers supported forced-labeling on every media type they were asked about, a more concerning result came out of an attached task for each respondent: determining whether a video was real or AI-generated.

The video-based survey was conducted using OpenAI's Sora; a text-to-video AI generator.

Participants were shown a total of eight videos, four which were AI-generated and another four that were human-made videos. A majority of viewers correctly guessed the origins of a video just once for the artificial-intelligence generations and once for the human-made videos.

While an AI video of a close-up on a person's eye had 50% of respondents declare it wasn't authentic, AI-content that showed panning footage of a town was correctly deemed fake by 56% of viewers.

A human-made video of a city was the only footage correctly labeled as created by a humans, with 57% of viewers saying of it was real.

The same survey respondents made it clear that they would also support government regulation in terms of protecting certain job sectors from the impact of artificial intelligence.

In total, 76% said that they would support the government implementing "strong regulations to protect jobs Sora and AI could impact." Just 24% said that "strong regulations will stifle innovation and prevent more jobs from being created by the new technology."

The survey of 1,082 U.S. adults was consistent across all demographics with those between ages 50-64 most likely to support the regulations and ages 35-49 least likely to support, at 81% and 71%, respectively.

Women were more likely to support the legislation by a factor of 6 points versus men.

Support for a few other notable regulation types were ranked highly by respondents. These included those who think there should be accountability rules for companies responsible for AI content output (39%), those who feel there should be stricter privacy laws for collecting user data (34%), and those who believe ethics standards for AI should be developed (33%).

Like Blaze News? Bypass the censors, sign up for our newsletters, and get stories like this direct to your inbox. Sign up here!