Openai is making its image generation software DALL-E more widely available to businesses with the launch of an application programming interface. It will be easier for companies to add DALL-E's text-to-image feature to their products, thanks to the availability of the application programming interface.
The Microsoft Designer app uses the software to create imagery for Office users, from PowerPoint slides to illustrations for homework. Microsoft is one of the major investors in Openai.
The company was excited to see the new applications that would be found for DALL-E.
Some customers are building on this in very interesting ways. Some of them are more business oriented. Miller gave an example of a startup called Mixtiles that is using the APIs to create posters and art for home decoration and another called CALA that is using it to help customers design their own clothing. He said it was always inspiring to see the ideas people came up with.
Rivals offer easier access to their software.
OpenAI, once the domain's clear leader, has been challenged by newcomers like Midjourney and Stability Artificial Intelligence. These organizations have made it easier for them to build on their artificial intelligence systems. Other players in this space, like Meta and Google, have taken a more cautious approach, developing systems with similar capabilities but limiting their public use to very limited situations.
There are many dangers associated with text-to-image artificial intelligence. There are ethical questions regarding data use and the software can be used to generate misinformation and harmful imagery.
Text-to-image systems like DALL-E use images from the web to train their systems. Many artists are angry that the resulting technology can not only be used to imitate their style but also that they have not been paid for the use of their work.
Some firms are starting to give compensation for their work. Shutterstock, for example, which licensed its contributor data to OpenAI to create DALL-E and which is using itsAPI to generate custom stock imagery, announced recently that it is setting up a Contributors Fund to reimburse individuals whose work is used to train artificial intelligence.
Miller said the company had nothing concrete in the works when asked if they were planning to compensate artists. Miller doesn't have anything specific to say on this right now. We are still looking for feedback from the community. It is a difficult question to think about from many different viewpoints. We would like to learn from the community.
The question of what training data was used to create DALL-E is not easy to answer, as the company has never shared what training data was used. According to legal experts, fair use doctrine in the US will likely apply to the training of artificial intelligence models. Legal cover isn't the same as ethical endorsement.
As the company spins up its systems, it will not be vetting customers in how they use the technology, according to Openai. DALL-E limits the creation of nudity, gore, and politically sensitive images. Customers will be able to choose between three levels of resolution. 512 x 512 images will cost $0.018 apiece, 512 x 512 images will cost $0.016 apiece, and1024 x 1536 images will cost $0.06 apiece.