In one video, 500 black-and-white photos feature the stoic look of Abraham Lincoln with his high cheekbones and bearded jawline.
None are real.
“All the images we see in these videos are generated by artificial intelligence,” said Abe Gibson, assistant professor of history at the University of Texas at San Antonio, who launched a new website called Honest Abe. posted the clip on. . “This is the new brave world we live in.”
On the one hand, these images can be educational. But it also raises questions about the acceleration and pervasiveness of the artificial intelligence used to create these works.
Many people may not know that there are only 130 confirmed snapshots of Lincoln, so if not for the video photo montage disclaimer, all the images contained therein are made using state-of-the-art apps. You may not have noticed that it was created by And that Lincoln’s famous speech (which was given in 1863 and was not in multiple languages like modern video) was given long before the first use of audio recording film. is known to most people, but technology applied in other contexts can easily fool unsuspecting people.
Related: San Antonio professors wrestle with AI to help students write, possibly cheating
Gibson is mostly part of the UTSA Department of History’s professorial team, which uses deepfakes (AI-generated videos, audio, and photos of real people doing or saying things they wouldn’t actually do). have created.
One afternoon in March, they created a Lincoln-inspired imagery using Lensa, Stable Diffusion, and Midjourney employing the new wave of viral AI generators. Another day, I used the AI face animator app Avatarify to develop a video of Lincoln giving his famous speech in English, Spanish, and several other languages.
“These tools can affect the historical record,” Gibson said in a telephone interview.
In the age of rapid advances in AI, more and more celebrity deepfakes (still images and videos, many with simulated audio) of celebrities are circulating on social media platforms like Instagram and Twitter.
But experts have expressed concern over technology that could generate pornography without the consent of Houston-area teachers. Fake video of Tesla, SpaceX and Twitter CEO Elon Musk speaking in a real estate marketing ad. And a song that mimics the voices of Drake and The Weeknd.
In 2019, Texas became the first state to ban the spread of manipulative deepfake videos intended to sway voters or harm candidates. But with few laws regulating the proliferation of technology, misinformation continues.
recently, AI-generated image of former President Donald Trump fighting police I caught a virus. The Republican National Committee produced a 30-second ad featuring a look-alike of President Joe Biden to address the fake city shutdown of San Francisco and other fictitious crises. Globally, there are deepfake videos showing Russian President Vladimir Putin warning Americans about election interference in English.
“AI-generated synthetic media will not only outstrip the amount of traditional human-generated content on the internet, but will ultimately end up being something we have been creating since the dawn of human culture thousands of years ago. more than all other media,” the professors write in their paper. “Honest Abe” website. “No wonder people are panicking. After all, if you can’t even trust your own eyes, who can you trust?”
Related: Texas becomes first state to ban political ‘deepfake’ videos
Meanwhile, in San Antonio, UTSA and other higher education institutions are promoting ChatGPT. ChatGPT is an app that allows you to solve math problems, imitate English conversation to explain concepts, and write simple but nuanced sentences.
Universities have operated without formal policies regarding the use of AI text and image generators.
Last year, Gibson hired UTSA History Associate Professors Cindy Armas and Jerry Gonzalez and others to develop the website as part of the Digital Literacy Accelerator Program through the U.S. Department of Education.
“It’s a stupid website, but it’s really an antidote because[deepfakes]have already done real harm,” Gibson said. “From what we learn and see, we see how they undermine trust in everything, and how it undermines our democratic institutions and all sorts of things. increase.”
Gibson aims to attract K-12 and higher education programs to its website. He hopes to apply for an education grant by the summer so the team can post lesson plans with suggestions on how to teach students about his AI-generated app, which is already in widespread use.
The professors are also brainstorming ways to use the tools in various projects that benefit communities in Southern Texas. Gonzalez, director of the UTSA Mexico Center, is looking at ways to use AI to create Spanish study guides to help people prepare for the U.S. citizenship exam. Rebecca Gonzalez, who is pursuing a master’s degree in history at UTSA, wants to recreate a famous political speech in Spanish.
“Technology itself is not evil,” Gibson said. “These tools can be used for educational reasons and practical purposes with a little creativity.”
In an ongoing project, the professors are using the ReFace app, which allows users to digitally paste their faces over celebrities’ faces, to recreate his beloved role in the Coen Brothers movie. Lincoln’s face was glued onto the body of Jeff Bridges, who played The Dude. The 1998 comedy “The Big Lebowski.”
In yet another experiment, they added Chat GPT-4, the latest OpenAI model released earlier this year, in the style of the late Toni Morrison and Ernest Hemingway, and singer-songwriter Taylor Swift. I asked him to rewrite the Gettysburg Address.
eric.killelea@express-news.net
