The number of available artificial intelligence text to image generators is rising rapidly, giving the public an array of new tools to create digital imagery.
One of the first tools, DALL·E studied the relationship between images and the text used to describe them. It uses a process called “diffusion”, which starts with a pattern of random dots and gradually changes this pattern towards the image when it recognizes certain aspects of this image. The original DALL-E was introduced in January 2021, and the tool, with its ability to turn any text description into a unique image, impressed not only experts in the field of artificial intelligence, but also the public. Since that time, other systems have already been developed that are capable of processing text into pictures. Access to other systems, such as Midjourney and Stable Diffusion, earlier was much easier for the average user. But now, since July 2022, access is open to all users.
Anyone can use this AI art generator – that’s the risk.
Studying the experience of real use is an important part of the responsible development and implementation of artificial intelligence. The company started with a preview of DALL·E 2 for a limited number of trusted users. As developers learned more about the capabilities and limitations of the technology and gained confidence in security systems, new users were gradually added and DALL·E became available in beta in July 2022.
The DALL-E 2 creators Content Policy does not allow users to create violent or political content. DALL-E output data passes through various filters that prevent the generation of images containing violence, pornography, as well as real faces and bodies of people.
A text-to-image program named Stable Diffusion that offers open-source, unfiltered image generation, that’s free to use for anyone with a decent computer and a little technical knowhow. It’s been embraced by the AI art community and decried by many traditional artists; it’s been picked apart, exalted, and worried over. Experts note that the ability of text—to-image conversion systems to create naked bodies without consent and realistic images of people is potentially dangerous – this is an excuse for propaganda, disinformation and much more. There are also bias issues. Since these systems are trained on images that are available on the Internet, they can reproduce unequal aspects of society. For example, if a few months ago the system was asked to draw a portrait of the CEO, it most likely gave out an image of a white man. But at the moment, with constant updates and training of the system, this should not happen.
OpenAI allows users of the AI art generator DALL-E program to edit images with human faces. Previously, this feature was unavailable due to concerns of misuse, but after improving filters to remove images containing sexual, political and violent content, access to this feature was opened. It will allow users to edit images in various ways. For example, you can upload a photo and create variations of the image, or edit certain functions, such as changing clothes. This feature will undoubtedly be useful to many users of the creative industries, from photographers to cinematographers.
The revolution in art is not something new, but this one, as some believe, may be final.
Some artists already feared that a new generation of artificial intelligence image generators could take their work away and spend the years spent learning their craft for free.
“This thing wants our work, it actively opposes artists,” wrote California-based film and game conceptual artist RJ Palmer in a tweet that was liked by more than 25,000 users.
In tweets, he stressed how well the results of artificial intelligence systems can imitate living artists. In one case he investigated, artificial intelligence even tried to reproduce artists’ signatures. However, I believe that no matter how accurate the generated picture is, a person will still appreciate something created by human hands more. And no matter how banal it may sound, there is much more value in the works made by the artist’s hands than in the generated picture on the web. After all, this is an individuality that, if artificial intelligence can acquire, it will take a very long time.
So, we cannot escape from progress in our world, no matter what fears there are, because everything is improving, and the society gets along with these innovations…
Sources:
https://www.theverge.com/2022/9/28/23376328/ai-art-image-generator-dall-e-access-waitlist-scrapped
https://openai.com/dall-e-2/#demos
https://www.bbc.com/news/technology-62788725.amp
https://www.washingtonpost.com/technology/interactive/2022/artificial-intelligence-images-dall-e/