How accurate is the Virus Tiktoc AI POV Lab History Video?

Imagine waking up on the streets of London in Rome two thousand years ago, during the Nile River in ancient Egypt or during the Black Death in the 1300s.

Over the past few weeks, AI-generated videos showing the perspectives of people awakened to various historical periods have become a virus in Tiktok.

Dan and Hogne are creators behind two accounts: POV Lab and Time Traveler POV.

British-based Dan told the BBC he was making these videos, “The idea of ​​looking at the past through a first-person perspective felt like a unique way to bring history back to life.” It’s there.

Hogne, 27, from Norway, adds that his video teaches people “thinking about the cool parts of history, helping them learn something new.”

Despite the video offering millions of windows into history, do many historians share concerns about the accuracy of the content and whether AI can truly revive the past? Or is the history of a sophisticated modernized version designed for engagement?

Muddy and misty streets, coughing townspeople, and distant victims of Plague Doctor’s Bell are all traits of Hogne’s most watched videos, which have earned 53 million views.

It has sparked appeal among many, but historian Dr. Amy Boenington described the medieval-themed video as “amateur” and “stimulating and sensational” rather than historically accurate. I’m doing it.

“It looks like something from a video game, because it shows a world that is meant to look real, but is actually fake.”

She points to inaccuracies like depictions of the large windowed house running through towns that did not exist in the 1300s.

Historian and archaeologist Dr. Hannah Platz also noticed significant inaccuracies in a video depicting the eruption of Mount Vesuvius in Pompeii.

“Because of the accounts of witnesses of the Pliny eruption, we don’t start with lava spitting out from anywhere, so it feels cheap and lazy to not use the wealth of historical information available to us.”

She also finds a small mistake – the stem-stemed wine glasses and pepper mills were not present, and people recline when they are not sitting in the meal chair.

“The bread rolls in the video are modern breads and given that they actually have carbonated bread, it’s really a shame that the people making the video have done some research and didn’t include that. ”

Dan, who created the Pompeii video, says he recognizes that many of the details in his video are historically inaccurate.

“The content generated by AI is not perfect. I strive for accuracy, but these videos are about evoking a sense of period, not 100% de facto recreation.

“They’re more of an artistic interpretation than a strict documentary.”

But Dr. Boeington is concerned about the impact these artistic interpretations will have on the history of rewrites.

“It’s extremely dangerous because people can manipulate history. For example, someone can create a video generated to AI that backs up Holocaust deniers.”

Most people know that the content of the video is not authentic, but the concern is “young people who learn about historical periods for the first time through these videos.”

Dan rejects these claims, saying his video is “not intended to be considered pure historical facts.”

“If viewers are interested, they recommend studying history yourself. These videos are seen as a way to stimulate curiosity about the past, rather than replacing actual history education. .”

Hogne says she feels “responsible” when creating these videos, and focuses on trying to make the videos as accurate as possible.

Dr. Platz is worried that misinformation has not been checked, and some viewers of the comments seem to be unaware that the video generated by the AI ​​is not based on historical facts. It’s pointing out.

“We see a lot of students using AI right now. The problem is that they see something like this and it’s reflected in us as if it were true. It’s about being there.”

Dan ensures that all videos are labelled as AI-created, and Hogne has said that misinformation existed long before AI, and that “people need to think critically about everything they see.” There is.”

All historians the BBC speak to have spoken to agree that Dan and Hogne’s videos have merit.

Dr. Boyington says that you can “inspire someone to “gate the history and do your own research,” but Egyptology professor Elizabeth Frood said, “If done accurately and surely, This is a great benefit for the masses. It raises the interest and awareness of history.”

Barbara Keys, a professor of American history at Durham University, saw an AI-generated video of someone working at the Chernobyl nuclear power plant on April 6, 1986, when the reactor exploded.

She sees potential benefits in these videos, but calls the clip a “black box” because there is no source information or transparency about what information was provided to the AI ​​program.

She says the concern about the inaccuracy of the video is that the retractor is based on post-disaster images rather than previous ones.

“This makes me think that Soviet technology is really bad when Soviet technology was actually very sophisticated.”

She believes the viral nature of these videos is not about their historical content, but about the public’s appeal to the AI ​​itself.

“There’s nothing very interesting about the video. There’s no information about the accident or what happened after that, so what AI can do is do.”

Professor Food is concerned that the videos are not accurate or reliably produced, and emphasizes that historical reconstructions should be based on extensive research and verifiable sources.

“To make this video, you need to be critical of this video because you don’t know anything about making it or what kind of information was provided to the AI.”

She commented in particular in a video about being a child in Egypt in 1250 BC.

The first thing that hit her was the tendency for the video to “homogenize complex ancient worlds.”

“Egypt has been around for thousands of years and people may not realize that this is a single snapshot,” she explains.

Overall, she pointed to an example of the video appearing to be “understudied” and the teacher in a school scene reading hieroglyphs backwards when it should read from right to left.

Hogne, who created the video, says he recognizes that his video has historical inaccuracies.

“I try to find them where I can, but I’m not an expert in history, so I don’t always see the error, especially when it’s a small thing,” he adds. .

Asked about the lack of transparency in the source material, Hogne says in the future he might consider adding a link to where he got his information.

“I think it’s cool to use the right source to make it completely accurate, but you’re the only one making these videos, and it takes eight hours to make each one.”

Hogne relies on Chat GPT to investigate periods and seek information about how people and places looked. He then generates the first image of someone looking at the landscape, from which “it’s a lot of interaction to make it right.”

Dan says it takes four hours to create each video, and says, “Use a variety of AI tools to generate, animate, and create realistic sounds.”

He watches the video, reads the documentation, and tries to try “historical accuracy” before making the video.

“From ‘traditional Italian robes’ to ‘cobblestone floors’, you need to provide all the details to the AI.