We live through images. Our brain processes images, photos, drawings better than letters and words. That’s how it’s designed. That is why our digital world is filled daily with new images appearing in Twitter posts, in Facebook posts, we communicate on Instagram, Tiktok and YouTube which are pure images.
Once again, OpenAI, the company I mentioned earlier, is developing an image generator with artificial intelligence. That’s right, creating images without human help, completely with the computer.
The program called: Dall-e2, baptized by the media as: Dalle-2 (referring to Disney’s Wall-E character) is the development of AI to generate images, to understand it better, let’s analyze some of its features :
The first is that it can create new images from the description of a text, for example if we ask it: “Canadian forest”, it will search for photos, paintings with this motif and it will show us a forest of pines or oyameles. But if we ask: “Canadian forest on Mars” you can combine: forest + the planet Mars. It seems impossible for a human, but Dall-e2 will present us with its own version of this or several images with the combination of both elements.
Second, the editing feature is impressive. As we see in some mobile phone photography apps that add faces, gestures and colors, in the case of Dall-e2 you can add or remove elements to a photo from shadows, reflections, textures according to the parameters that let us specify. For example, if we want to reduce the colors or increase a shadow in a non-existent place, the program will do it without any problem, adjusting the image and the rest harmoniously without the change causing errors.
A third feature is the ability to enlarge images of the original canvas. OpenAI on its site shows an image of a classic painting, of a farmer’s wife and how Dall-e2 enlarges the shelves around it with pots from that period, expanding the painting that was just a fragment captured by the artist, but that the AI recreates from its perspective or with the parameters we gave it.
A fourth function is that you can take an image and create different variations of it, inspired by the original. Don’t hesitate to find a Mona Lisa of Asian descent, with light hair and an urban background. Or see a picture of Gandhi rejuvenated and with Western features. Both would be the work of Dall-e2.
The fact that an AI can take images, modify them, and recreate classic works of art opens up a field of possibilities and doubts that need to be considered. This technology, available on a trial basis, allows us to change our relationship with the images, to question whether they are true or false or whether they have been modified by the computer; In itself, this breakthrough in artificial intelligence leads us to question whether everything we see online is real. The subject forces me to do several reflections that I have to do in a next collaboration.