The exponential evolution of AI continues full steam in 2021. This week, OpenAI, a research startup focused on building the first general artificial intelligence, released another very impressive piece of AI called DALL-E.
The new algorithm is capable of creating any kind of image or illustration directly from a text description. It is something that, up to now, used to belong to science fiction.
The user just needs to enter a text prompt and the AI will take care of the rest. It understands the words and context and generates pictures, drawings, or illustrations of what you have just written. To be clear, this is not a bunch of images sitting in a database somewhere that are pulled according to a specific tag. These images are created in real-time and do not exist in the real world.
In the example below, we've entered: “a T-shirt with the word Skynet written on it.” DALL-E returned the samples below.
On this one, the prompt is: “a clock in the form of a Pikachu.” Look at the incredible results! Again, you can reverse search these images on Google and you won't find them anywhere on the Internet.
Another example: “a … of an Eagle sitting on a mountain at Twilight.” In this example, the AI was able to generate the images using different styles (replace the … with the words in the description). It is absolutely incredible.
Why should you care?
We've just mentioned in our December newsletter about fake human faces that, soon, everything in the world could be created by artificial intelligence in real-time. It seems the technology got here faster than we thought and it can even combine abstract concepts such as “a pig with the texture of a pie.”
OpenAI still hasn't released the technology publicly as they need to understand its impacts on the world. On one hand, I see this rendering the work of artists, photographers, and illustrators completely moot. On the other hand, it will empower any person to create anything they want for free and in seconds. It will help us to create faster storyboards, wallpapers, social media posts, or even business presentations.
If you are in the field of visual arts, you may want to learn how to program. Your work likely will be as good as the parameters and calibration behind the AI algorithms.