AI-produced art is subtly starting to change culture. The ability of machine learning algorithms to produce imagery from text prompts has drastically improved in quality, accuracy, and expression over the previous few years. These technologies are now leaving research facilities and entering the hands of regular people, where they are fostering the development of new visual languages of expression and, most likely, new kinds of problems.
Only a small number of top-tier image-generating AI are thought to exist at this time. They are difficult and expensive to make because they need access to the millions of photographs that are necessary to train the system (which duplicates patterns in the images it finds) and a lot of processing power, which may reach into the millions of dollars in cost.
When the output of these systems is splattered across the cover of a magazine or used to create memes, it is now primarily treated as novelty. But right now, designers and artists are incorporating this software into their work processes, and soon, AI-generated and AI-augmented art will be commonplace. Copyright concerns (who owns the image? Who created it?) as well as concerning potential threats (such as biased output or AI-generated false information) will need to be addressed right away.
Read more about this at thevege.com
