OpenAI Unveils DALL-E 3: A Game-Changer in Text-to-Image AI

Politics, News Sept. 21, 2023, 9:46 a.m.

OpenAI is launching DALL-E 3, an improved text-to-image AI with better understanding of user intent, enhanced image generation, and improved security. Its set for a phased release starting next month.

OpenAI is set to introduce the third iteration of DALL-E, its text-to-image AI system, bringing significant enhancements over its predecessor. Notably, DALL-E 3 will seamlessly integrate with the ChatGPT app, allowing users to harness its image generation capabilities. OpenAI demonstrated this integration to The Verge, showcasing how users can instruct ChatGPT to compose detailed prompts for DALL-E to work with.

OpenAI emphasized that DALL-E 3 exhibits marked improvements in understanding user intentions, particularly when presented with intricate and lengthy prompts. In cases where users struggle to articulate their requirements effectively, ChatGPT can assist by composing comprehensive prompts. During the demonstration for The Verge, DALL-E successfully generated four distinct images in response to a prompt requesting a ramen restaurant logo set in mountainous terrain.

OpenAI has taken steps to mitigate potential legal issues. While the current version of DALL-E can create images in the style of living artists, the upcoming version will decline requests to mimic their work. Artists will have the option to submit their own work through a dedicated form on the website, requesting its removal if necessary.

DALL-E 3 is scheduled for release next month, initially exclusive to ChatGPT Plus and enterprise customers. Subsequently, it will become accessible to OpenAI's research labs and API customers in the upcoming fall season. However, the company has not yet disclosed a specific date for its general public availability.