Over the weekend, an image of Pope Francis in a bulky down jacket hit Twitter, and it quickly went viral as users believed it was real. In fact, the image was first posted on Reddit along with three other images in the same style and are the result of the Midjourney AI generator.
If you look closely, you can see that the artificial intelligence did not do very well with the Pope’s right hand.
AI image generators like Midjourney, which also include OpenAI’s DALL-E and Stable Diffusion, allow you to generate an image simply by describing it with text. This became a reality because neural networks were trained on literally millions of images from the Internet, including many pictures protected by copyright. This has already become the reason for lawsuits from companies such as Getty Images, which claim that the developers of the Stable Diffusion AI generator have infringed their copyrights.
And while Getty has a good chance of winning this lawsuit, it’s unlikely to stop the development of AI-powered image generators. Midjourney V5 in its alpha version is already much better able to generate hands, and its capabilities may improve even more before release.
Interestingly, the US Copyright Office recently announced that AI-generated images cannot be copyrighted, so they can be used completely freely, including in books and magazines.
It’s also worth noting that AI-generated images are relatively new, with developers introducing their tools only in the past four months. But fake photos have been around for a long time, and only time will tell if AI generators will create more misinformation than Adobe Photoshop once did.
Loading comments …