Three ways to avoid being fooled by AI slop

by Silvia Montana Nino, University of Melbourne and T. J. Thompson, RMIT University

The global community produces and uploads billions of images Hundreds of thousands of hours of video On the Internet every day.

The problem is that some of this content is misleading or downright wrong. And when it’s in visual form, it can be Particularly convincing.

Take, for example, the Met Gala held earlier this month in New York. While paparazzi were snapping photos of Rihanna, Beyoncé, and Nicole Kidman strutting their stuff, others saw “photos” of celebrities, like Rosalía, Lady Gaga, and Jacob Elordi, who were actually elsewhere (pics below). Instagram carousel Artificial intelligence is created).

While this kind of AI nonsense may seem harmless and easily verifiable, other “media fakeouts” are becoming more problematic and demanding. More powerful technologies To verify.

Traditional verification techniques are becoming insufficient as AI becomes increasingly persuasive and the line between authenticity and artificiality blurs. This applies to all content, from still images to moving images and audio deepfake.

The volume of content and the speed at which it travels doesn’t help. It also doesn’t help that fact checking can take hours or days while fake products can be created in seconds.

First, prepare yourself

Evidence When discovering AI-generated content, it suggests multiple strategies and acknowledges that there are no perfect solutions. But there are helpful things you can do.

Get to know yourself with Examples of counterfeit products And study how to verify the validity of these facts. This helps you understand what’s possible and learn how fact-checkers sort out the truth from the fake.

Look deeply. Zoom in. Pause the content or watch it frame by frame. Examine the small details. Look for inconsistencies, flat textures when they shouldn’t be, or patterns that are too perfect or incomprehensible. Does the location shown match where the scene is supposed to be? Do the shadows fall naturally and do the lines follow the rules of perspective?

See broadly. Are you familiar with the source? What else is he posting and how long has it been? What do other reliable sources say? How does this imaging compare to others available? Or if there are no others available, should that make you stop?

Then apply what you’ve learned

Let’s take an example and work through it together.

this Facebook reelposted by an account called “Real Talk Hub”, aims to show migrants being stopped and turned back by Australian police at the airport.

Before we get too deep, let’s evaluate the opening image.

The video uses a scale to show what appears to be a long flow of passengers. Some move toward and some move away from the plane. It’s hard to make out details in the video. Overlaid text obscures almost all of the horizon line. A shallow depth of field makes distant aspects blurry and difficult to distinguish.

Many passengers have dark skin and are visually coded as “other.” They interact with a light-skinned police officer who is taking notes on a clipboard.

The vertical video is carefully framed so as not to reveal identifiers such as the airline name that appears to begin with the letter “P.” This makes it difficult to research the name of the airline and whether reliable sources support the story being told.

Although the people and scenes appear realistic at first glance, the integrity of the video is revealed when we slow down and look closer. People in the passenger line shift and turn.

The officer is able to remove the paper from the clipboard on his own and it appears to inexplicably leave behind white lines. The police jackets look different from the pictures You can find verified media images Australian Federal Police.

All these clues together suggest that the video was created by artificial intelligence.

Think like a fact checker

Many AI-generated videos can trick you and create a very convincing story. Therefore, fact-checkers have developed triangulating methodologies that examine elements beyond just what you see on the video.

One way to do this is to systematically investigate contextual factors – other things surrounding the content. our Team research I’ve found that professional fact-checkers are usually interested in what type of social media accounts or websites are distributing questionable media.

For this AAP Verified In a video about banning dogs on the beach, it was necessary to examine user activity and posting patterns.

In addition to visual distortions, fact-checkers also found Invisible watermark Which helped them determine that the content was created by artificial intelligence.

@aapfactcheck

A foreign-run Facebook page is using videos created by Al to stir up dog owners’ anger against Muslims in Australia. Click the link in our bio to read the full report.

♬ Original Sound – AAP FactCheck

Other things to check are how long the social media account has been running, how often the social media account posts, and whether the account is transparent about its use of AI.

However, these are not reliable indicators of authenticity. The immigrant example above comes from an account that is about five years old. It also comes from a “verified” account, which may make it appear more credible. But both Facebook and X Now let users pay for this verification.

In general, when it comes to suspicious photos or videos, don’t just look deep. See also Widely.

Content generated by artificial intelligence Our eyes can increasingly deceive usso you also have to look beyond what’s in the video. A multi-method approach that takes into account visual and contextual clues can help. By training your ability to think like a fact-checker, you can stay safer online.

Silvia Montana NinoLecturer at the Journalism Development Center, University of Melbourne and T. J. ThompsonAssociate Professor of Visual Communication and Digital Media, RMIT University

This article was republished from Conversation Under Creative Commons license. Read Original article.