Can you still distinguish between real photo and AI image? Prove yourself in the test

With artificial intelligence (AI) like Midjourney or Microsoft’s Bing Image Creator, even laymen can create any image in no time. A short text description such as “draw a picture of rabbits taking selfies together” yields these four AI-generated images after a few seconds.

Three of the following photos are real and one is made by artificial intelligence. Do you recognize the AI ​​image?

What image has been created by an artificial intelligence?

Which dog is fake?

Which child doesn’t even exist?

Which piece of meat is generated by AI?

Which woman only exists virtually?

Which tiger should no one be afraid of?

In which apartment can’t anyone put their feet up on the sofa?

Which chalet will you never find in the Swiss Alps?

Which Matterhorn Did Artificial Intelligence Create?

All real photos in the article are courtesy of Shutterstock photography agency. The AI-generated photos were created using the current version 5 of the text-to-image generator Midjourney.

Version 5 is only available in an early alpha version, but some results are already so good they’re barely distinguishable from photos. This is especially true for landscape, food and nature shots. Problems are reported by AI image generators with eyes and hands and objects held by the hands. These are sometimes strangely distorted. With Midjourney in the latest version, this is not perfect either, but it has improved enormously.

Professional photographers and artists can use AI tools to create photos and paintings that are much more realistic or creative than our first attempts at Midjourney. A selection of impressive examples can be found here.

If you want to become an AI artist yourself, you can get started with the following text-to-image generators: