Is Seeing Believing? AI Videos Look Extremely Real

Over the last year or so, the quality of AI-generated videos has become so good that it’s extremely difficult to tell whether a video is real or not. An English TV channel recently ran a show about AI. At the end of the episode, the “woman” presenting the show announced that she wasn’t real.

Deep Learning and Generative AI
     AI tools are usually trained using “deep learning”. In deep learning, computer programs sort deeply through huge amounts of information, allowing them to find patterns that humans often miss. These patterns get stored in a “model”, which can then be used to apply those patterns in new and sometimes surprising ways. Generative AI uses these patterns to create new text, images, music, videos, or other media that didn’t exist before.

Over the last several years, large tech companies have been pouring lots of effort into developing generative AI models. These are models that can create new works based on a short suggestion, called a prompt. The amazing power of these tools has raised concerns that many workers could lose their jobs to AI in the future.

The most famous of these tools is probably OpenAI’s ChatGPT, which can very quickly write computer code, research papers, poems, and songs. It can also answer challenging questions, and do complex math. Now many large tech companies have their own versions of these text-generating models. And they are being used in all kinds of places. Google has even replaced its regular search tool with an AI model.

AI models have been trained to create much more than just text. They can create music, artworks, realistic photos, and even videos. At first, the work created by these tools was impressive, but it often had obvious problems, such as a person with six fingers.

But with time and lots of training, most of these problems have now been fixed. Several companies now offer cheap or free tools to create short realistic videos. These tools can now create images and videos so realistic that they can’t be told apart from real pictures or videos.

This can create problems. After last week’s Hurricane Melissa, fake images and videos spread quickly on social media. One video falsely suggested that the hurricane had blown sharks into local swimming pools.

Many AI videos are intended to be funny. But it’s easy to see how the tools could be misused, and could lead many people to believe things that are not true. This could be especially powerful and dangerous in politics.

TV Show on AI Has Surprise AI Presenter

On October 20, the English news show Dispatches ran a special episode called “Will AI Take My Job?” The show aired on the TV network Channel 4. In it, a reporter named Aisha Gaban discussed how AI could affect many different kinds of jobs in the future.

At the end of the show, the presenter revealed that she wasn’t real, but was actually an AI creation. “I don’t exist,” she said. “I wasn’t on location reporting this story. My image and voice were generated using AI.” Channel 4 said the show was meant as a reminder of the power of AI and “how easy it is to hoodwink [trick] audiences” when they have no way of knowing whether a video is real or not.

The Channel 4 show comes soon after the news of a new AI “actress” called Tilly Norwood. The company behind the AI-generated character says AI actors could save movie companies lots of money. The announcement has upset many people in the movie business.

Did You Know…?
Currently, there aren’t any perfect methods for figuring out if a video is made by AI or not. Some companies put a mark on their videos to show it was made by AI. But the marks can be removed. If you aren’t sure you can trust the source of a video, it’s best to be cautious until you can confirm the information some other way.

Note: Beginning in 2023, NewsForKids.net began testing the use of LLMs, including products from OpenAI, in collating information and creating early drafts of some articles. This article, however, was 100% human-written from start to finish.