NUNZIUM

Progress in the domain of Artificial Intelligence (AI) is expected to change the world in a few decades. Such continued progress has been recently demonstrated by applying AI programs to games such as Chess and Go, where computers proved to be able to learn and win against the top human champions. Recently another AI program, DALL-E, demonstrated unprecedented power in creating original images. The program is now available online (see link below), and anyone can use it: it asks the user to enter the content of the desired image and, in a matter of seconds, creates images reflecting the user's requests. DALL-E is one of the most discussed, appreciated and criticized artificial intelligence algorithms, especially since the end of the summer when a trial version of it was made available to everyone. The first version of this artificial intelligence was released in January 2021, but with limited capabilities compared to the current ones, which surprised many observers and instilled concern in illustrators, graphic designers and artists. OpenAI carried out the development of DALL-E. This research laboratory is part of the OpenAI LP company, which is, in turn, controlled by Open AI Inc., a non-profit corporation. Billionaire Elon Musk founded the organization in 2015. For some months now, DALL-E has been the best-known system for producing images with algorithms, but there are others. Several other research groups, developers, companies, and organizations have made AI drawings, such as Midjourney, Imagen and DreamStudio. Each of these systems employs different algorithms but with similar operating principles. The original breakthrough came in 2016 when a team of scientists managed for the first time to ask an AI algorithm for the content of an image through a written request. The initially designed algorithms, however, were quite complex to use and specialized only in a small niche of demands. The most recent developments result in systems such as DALL-E that can practically design anything, with ever higher levels of adherence to requests. The latest generation of these AIs is straightforward to use. Just as researchers in Canada had experienced in 2016, but with much larger and more defined images, it is now possible to write a request in natural language and obtain the corresponding drawing in a few moments. Many aspects of DALL-E functioning are unknown, but its programmers have reported adding elements to ensure that the images produced respond to a specific aesthetic taste. As far as possible, the AI ​​produces images that we should like and, consequently, that we believe are more in line with our requests, amplifying that impression of having what we asked for. For example, asking for a specific desired painting style is possible. Those who deal with graphics and illustrations have begun to show a particular concern, sometimes intolerance, towards the AIs that draw and manage to draw better and better. One day, perhaps not too far away, for some types of illustrations, the work of professionals could be replaced by AI. Even if this did not happen in other sectors, such as writing - where there is massive room for improvement - the advances in drawing with AI have been faster and are, above all, more promising.