OpenAI Introduces DALL-E 3, Offers Artists the Option to Refrain from Participation in Training

OpenAI has introduced an enhanced version of its text-to-image tool, DALL-E, which leverages ChatGPT, the popular AI chatbot, to simplify the process of generating image prompts.

Cutting-edge AI-powered image generation tools typically rely on prompts—descriptions of desired images—to create artwork in various styles, spanning from photorealistic to fantastical. However, crafting the perfect prompt can be a challenging task, to the extent that “prompt engineering” has emerged as a recognized profession.

The latest iteration, DALL-E 3, employs ChatGPT to assist users in refining prompts. Subscribers to OpenAI’s premium ChatGPT plans, ChatGPT Plus and ChatGPT Enterprise, can now input a request for an image and fine-tune it through interactions with the chatbot. The results are directly provided within the chat application.

ChatGPT can take a brief prompt and enhance its descriptiveness, offering more guidance to the DALL-E 3 model.

The integration of ChatGPT is not the only update to DALL-E 3. According to OpenAI, it also generates higher-quality images that more accurately correspond to prompts, especially when dealing with longer and more complex prompts. Additionally, it handles content that has historically posed challenges to image-generating models, such as textual descriptions and depictions of human hands.

Furthermore, DALL-E 3 incorporates new mechanisms to reduce algorithmic bias and enhance safety. For example, it will decline requests for images mimicking the style of living artists or featuring public figures. Moreover, artists can now choose to exclude certain pieces or all of their artwork from being used to train future iterations of OpenAI’s text-to-image models. This move follows a lawsuit against OpenAI and some of its competitors, alleging the use of copyrighted artwork to train their generative AI image models.

The release of DALL-E 3 occurs as the competition in the field of generative AI, particularly in image synthesis, intensifies. Rivals like Midjourney and Stability AI continue to refine their image-generation models, placing pressure on OpenAI to remain at the forefront.

OpenAI intends to introduce DALL-E 3 to premium ChatGPT users in October, followed by research laboratories and its API customers. The company has not disclosed whether it plans to launch a free web tool, similar to what it did with DALL-E 2 and the original DALL-E model, and when it might do so.