Dalle-2 Is Available to Everyone
Recently, there’s been a lot of excitement about various image-generating algorithms, including MidJourney, Google Research’s Imagen, and Open AI’s Dalle-2. These are artificial intelligence systems that use a blend of natural language processing algorithms together with massive image training sets to come up with extremely impressive results. So, whether a person using the AI wants to see a painting of a fox in a field at sunrise painted in the style of Claude Monet or a majestic oil painting showing a raccoon Queen with a French royal gown, they’ll get results. The pictures that the AI produces can be very accurate and often surprise those who order them.
Dalle-2 Is Developed by OpenAI
OpenAI developed the first version of Dalle and let the public have fun with it in January 2021. They soon announced they were working on an improved version of the AI and revealed Dalle-2’s beta version in July 2022. Access was initially restricted to pre-selected users who could do a research preview of the system. This was necessary due to concerns about ethics and safety, but since 28 September 2022, Dalle-2 has been made available to the general public, and the experiments with it have started producing stunning results.
Despite its advanced algorithms, DALL-E 2 has its technical limitations when it comes to language understanding. It could fail to distinguish a red book and a yellow vase from a yellow book and a red vase or panda making latte art from latte art of a panda, but it still generates creative images, despite those being incorrect compared to the input. Still, the technology will certainly get even better as time progresses.