Simple question: Why can AI design images and videos for from text?

It is known that Chat GPT is able to get information from the network, but is it only text? How can AI interpret information from images? From the same AI research company where Chap GPT comes from (OpenAI), DALL-E 2 is another product that works with AI to create images from text. Actually, all of us network users are co-responsible for AI having such a creation capability, and one of those commands that AI recognizes is alt text or ALT, which is describing in words the content of a non-textual element.

For example, when posting on LinkedIn, whether it is a document, image or video, it is suggested to add an ALT description, which, in a nutshell, means to describe the medium briefly according to its content "as if you were describing it to a blind person". 

In this way, not only programmers and designers of web pages (and other digital platforms) have been contributing to the ALT description of media, but anyone can do this simple exercise in which we ourselves, as users and consumers, have been building the path by which AI is able to know what is in an image or video.

Whether it's for a job, for your resume or any project that requires a graphic description of a medium, you should save that file under a name that describes its content with a maximum of four keywords separated by a hyphen.

For example, think of three keywords describing the following image:

We can call it: toy-didactic-forms.

And for the alt text or ALT, how would you describe it? Let's do the exercise again, how about... educational toy with geometric shapes?

So it is that, in part, over the last few decades, artificial intelligence has taken our descriptions from alt-text to be able to create its own graphical representations when requested to do so.

If you find any figure on the internet that doesn't have this description in the file name, surely the artificial intelligence is discarding it, although its recognition ability has been evolving in a scary way.

Try DALL-E and let me know what you think.

_________

By: Sandra Hernandez