DALL-E 3 is official. OpenAI just announced the third version of its artificial intelligence to generate images from text, which comes with several improvements and new features. The most interesting thing is that the technology now natively integrated into ChatGPTallowing users to create better prompts directly from the chatbot.
According to Sam Altman’s startup, DALL-E 3 evolves considerably compared to its predecessor when it comes to interpreting user requests. But this is where the closer contact with ChatGPT also stands out.
When producing an image, It will be possible to ask the chatbot to generate it, or to create the prompt in question. ChatGPT will provide a longer and more detailed indication that will allow DALL-E 3 to better interpret it and provide more accurate results.
“Modern text-to-image conversion systems tend to ignore words or descriptions, forcing users to learn prompts. “DALL·E 3 represents an evolution in our ability to generate images that adhere exactly to the text you provide,” says OpenAI.
OpenAI has made it clear that the qualitative leap in the images generated with DALL-E 3 is considerable. The new technology, the company claims, understands better how to compose a scene and what is the connection between the elements that are part of it.
DALL-E 3 promises better results thanks to its integration with ChatGPT
DALL-E 3 promises better results when including text within drawings, as well as when dealing with parts of the human body that it previously failed to interpret correctly. For example, the hands. This was one of the AI’s weakest points when comparing its creations with those of other platforms, such as Midjourney.
However, the developers have not focused DALL-E 3 only on improving the quality of the results or offering a better interpretation of the results. prompts. They have also implemented more robust security measures in order to mitigate bias and prevent the use of the tool to create potentially harmful content such as deepfakes.
It is because of that DALL-E 3 will refuse to create images of public figures based on their name. And due to growing concerns about the spread of counterfeit content, it will also not offer results based on the style of living artists. What’s more, creators will be able to request that their work not be used to train the artificial intelligence model that powers this tool. Although the changes would be applied in future versions of the platform.
On the other hand, Sam Altman’s team reported that they are researching how to help people detect images created with artificial intelligence. The company is already testing an internal tool, but there has been no mention of whether it is planned to be released publicly.
If you want to try DALL-E 3, you will have to be patient. The new version of generative AI will arrive in October to ChatGPT Plus and Enterprise subscribers through its API. Research laboratories, meanwhile, will have access starting in the fall. For now, OpenAI has not announced when it could be available to the general public.