Microsoft said it would soon bring AI-powered art tools to its Bing search engine, Edge web browser, and other programs. The AI will be part of the company’s new Microsoft Designer program. The advanced DALL-E 2 algorithm will power the art software. “The biggest benefit of AI-generated art is that it helps democratize creativity—enabling everyone with the power to generate visual art,” Alon David, the vice president of product for the popular app Facetune, which uses AI to manipulate images, told Lifewire in an email interview. “It presents tools that empower anyone to create images that are creative or inspiring without needing to spend years mastering tools and techniques that professional artists use.”
Smart Art
Companies like OpenAI and Meta have recently released tools that can generate a picture based on a given text, Rishabh Misra, a machine learning engineer at Twitter, told Lifewire in an email. Such tools are trained to learn the correspondence between a picture and a text from image and caption pairs publicly available on the internet. “For example: if we supply such AI tools with enough examples of dog pictures with the word ‘dog’ in the caption, they can learn a visual representation of a dog,” he said. “This learning is done on a large scale for various entities. Due to this reason, such AI tools are able to curate artworks with many distinct characteristics provided in the text.” You might want to try AI-generated art tools to spark creativity. Tech consultant and IEEE Member Yale Fox told Lifewire via email that there are many times when “I have had an idea for an image in my head that I just can’t get down on canvas. The AIs can do an excellent job in a few seconds for the cost of a few pennies.” AI art isn’t limited to a visual medium. Yale noted that blogs no longer have to use generic stock art and can instead generate an image that is exactly how the user wants, but also stylistically on brand. Similar models are in development now that can generate a screenplay and then create a movie or virtual reality experience from this text.
Art Gets the AI Treatment
AI-generated art seems like it’s coming from thin air, but it’s actually thanks to the power of algorithms. Fox said that the software works when a user enters some descriptive text (called a prompt) for the image they want to generate. “For example, ‘a dachshund dressed up like Santa Claus’ will result in the placement of these items on a canvas,” Fox said. More and more noise is added to the image until it is not recognizable. Then, Fox said, the algorithm now knows how it can go from start to finish, and so it also knows how to go in reverse. When given a new sample of noise, it works backward and denoises the image in a similar pattern to the forward process. “In a way that is stranger than science fiction, a new image is generated,” he said. Not everyone is a fan of AI-generated art. Rich Hilliard, a professor of graphic design at James Madison University, told Lifewire in an email interview that AI art programs “are incredibly fast in execution, but there is a certain stylistic ‘sameness.’” Hilliard also is concerned about whether using AI tools will reduce human creativity. “The reliance on machine intelligence reduces the artist to ’technician’ although, after generation, an artist can refine the image themselves,” he added. “Unfortunately, this really diminishes the unique aspects of individual image creation.” But Misra noted that although AI-generated artworks look realistic and are easy to produce, generating appealing images still requires creative expertise. “Thus, AI art can be seen as a stepping stone to generating more creative and complex artworks by human artists,” he said. “The future might have human artists guiding AI tools based on their creative concepts to generate artworks that were not feasible before.”
