Quick News Bit

Animals that are too cute to be true: how to detect AI-generated images

0

Whether it is a baby sloth hanging onto someone’s thumb, a tiny colourful peacock or baby penguins taking a selfie, these insanely cute images of animals have been shared thousands of times on social media in recent months by people unable to resist. But it turns out that all three of these images were generated by AI. We’ve written up a list of tips so that you won’t be duped just because something is wildly cute.

Issued on:

6 min

What do these three images have in common? They’ve all gone viral since early May, for one. They also all feature tiny, unbelievably cute animals. 

This image of a baby sloth cupped in someone’s hand, for example, garnered more than 265,000 views and was shared more than 3,000 times on Twitter.

The image of this cute baby sloth was actually generated using artificial intelligence.
The image of this cute baby sloth was actually generated using artificial intelligence. © Twitter

On Instagram, a photo of a tiny baby peacock has garnered more than 5,800 “likes” since April 28, when it was posted by an account called “Beautiful nature”.

This teeny, tiny peacock is a fake image generated by artificial intelligence – if you couldn’t tell.
This teeny, tiny peacock is a fake image generated by artificial intelligence – if you couldn’t tell. © Instagram / Birdlovers_

However, the image that got the most engagement online is one of baby penguins seemingly taking a selfie. Since it was posted on July 7, it has garnered more than 46 million views on Twitter.

This image of baby penguins garnered more than 40 million views … turns out, it was generated by artificial intelligence.
This image of baby penguins garnered more than 40 million views … turns out, it was generated by artificial intelligence. © Twitter / @shouldhaveanima

These three images have more in common than just being cute they were all generated by artificial intelligence. That means these aren’t real photos, which obviously means that these animals aren’t real either. 

How plain old observation is often your best tool for spotting fakes 

Janne Ahlberg founder of the site Hoaxeye, which identifies fake images circulating on social media took a look at these three images. 

Ahlberg told our team that the easiest way to spot an image generated by artificial intelligence is to look for “artifacts” or issues with the digital image. Essentially, the tools that generate images using artificial intelligence don’t always get it right and often leave behind errors.

Take a look at the baby sloth picture, for instance. There are two errors that stand out right away. First of all, the thumbnail has an issue  it looks like it has been cut in two. There is also something going on with the bottom left of the image. 

Also, there’s a factual error. Sloths only have two or three fingers, which isn’t the case with the baby animal in this picture. 

As you can see, the thumbnail looks a bit dodgy – one of our first clues that this might have been generated using artificial intelligence.
As you can see, the thumbnail looks a bit dodgy – one of our first clues that this might have been generated using artificial intelligence. © Observers

Something else gives away the image of the tiny peacock. There are two points of focus in the image  the little bird’s head and its feet. The rest of the image is blurry. It’s not possible for a real photo to have two different focus points. 

And, again, there is also a factual error here, too: baby peacocks are actually usually brown or gray and, to be honest, pretty ungainly looking. It takes them a few years to develop their colourful feathers. 

There are two points of focus in this image – the peacock’s head and its feet – while the rest of the image is blurry. A photograph can’t actually have two points of focus.
There are two points of focus in this image – the peacock’s head and its feet – while the rest of the image is blurry. A photograph can’t actually have two points of focus. © Observers

As for the little penguins, if you take a closer look, you’ll see that something isn’t quite right with one of the birds in the background  it seems to be made up of two images that aren’t perfectly aligned. This is a sign that the artificial intelligence wasn’t able to generate this part of the image. 

You can see an error with one of the penguins in the background – the image of its head isn’t aligned.
You can see an error with one of the penguins in the background – the image of its head isn’t aligned. © Observers

A tool to help detect fakes but use with care 

There are several online tools that you can use to detect images generated by artificial intelligence. Our favorite as of July 2023 is Hive Moderation. If you upload an image to the site, it will give you a percentage indicating how likely it is that the image was generated by artificial intelligence. 

We ran these three photos through Hive Moderation, which told us that there was between a 99.6 percent and 99.9 percent likelihood that the images were generated by artificial intelligence. Good call, HiveMod.

A tool called Hive Moderation concluded that these three images were generated by artificial intelligence.
A tool called Hive Moderation concluded that these three images were generated by artificial intelligence. © Hive Moderation / Observers

The site claims that it is 99 percent accurate in analysing if images are AI-generated or not. However, like any algorithm, this tool learns from the images fed into it and, thus, can be easily fooled. 

A number of social media users, many of whom were based in Japan, proved they were able to fool the tool. It gave very different responses if the images were modified even in a minor way. Thus, it can’t be considered a foolproof tool for analysis.

The photo on the upper left is real but has been modified with the app FaceApp. Hive Moderation, however, concluded that it was generated by artificial intelligence. The same photo, on the bottom right, has been slightly altered in a different way and, bizarrely, the tool no longer says it was generated by artificial intelligence.
The photo on the upper left is real but has been modified with the app FaceApp. Hive Moderation, however, concluded that it was generated by artificial intelligence. The same photo, on the bottom right, has been slightly altered in a different way and, bizarrely, the tool no longer says it was generated by artificial intelligence. © ken5bt

So why are people sharing fake cute animal images?

Everyone on the internet or nearly everyone seems to love a good animal photo. But that doesn’t explain why people are trying to pass off images of cute animals generated by AI as the real deal.

Our team asked expert Janne Alhberg of the site HoaxEye

I guess that, first of all, these cute animals have gotten popular simply because people seem to like them and because they work, a lot of people copy the technique. 

The accounts that share AI-generated images trying to pass them off as real have different goals. But a lot of them just want to get a lot of likes and followers. 

Some accounts are more financially motivated. Some of the accounts are hoping to be sold off once they have a lot of followers [editor’s note: on the black market, people will pay hundreds of dollars to get thousands of followers].

This is a screengrab of a website where it is possible to buy several thousand followers for about a hundred dollars. Our team blurred out the names of the accounts.
This is a screengrab of a website where it is possible to buy several thousand followers for about a hundred dollars. Our team blurred out the names of the accounts. © Observers

Back in January 2018, we interviewed Ahlberg for another article (see below). Back then, we asked her why so many accounts were trying to pass off fake videos of incredible natural phenomena as real. In 2023, the subject might be different, but the idea remains the same.  

For more on this topicDebunked: When amazing nature shots are a bit too good to be true

For all the latest Health News Click Here 

 For the latest news and updates, follow us on Google News

Read original article here

Denial of responsibility! NewsBit.us is an automatic aggregator around the global media. All the content are available free on Internet. We have just arranged it in one platform for educational purpose only. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials on our website, please contact us by email – [email protected]. The content will be deleted within 24 hours.

Leave a comment