OpenAI's ChatGPT will let users generate images with DALL-E 3 integration

OpenAI has baked the updated DALL-E 3 into ChatGPT, allowing users to generate images with simple prompts. (Image Source: OpenAI)

Artificial intelligence startup OpenAI has debuted a new version of its DALL-E image generator and integrated it into its popular chatbot ChatGPT on September 20. The move comes as the Microsoft-backed firm looks to capitalise on ChatGPT's popularity among consumers to maintain its dominance in the nascent generative artificial intelligence space amidst rising competition from tech giants such as Google and upstarts such as Midjourney and Stability AI.

Called DALL-E 3, OpenAI claims that the image generator can understand significantly more nuance and detail than the company's previous systems, particularly complex prompts, to create "detailed and accurate" images.

"DALL-E 3 represents a leap forward in our ability to generate images that exactly adhere to the text you provide" the Sam Altman-led firm said in a blogpost.

The startup said the new image generator also delivers significant improvements in generating text within an image and in human details such as hands, as compared to the previous version DALL-E 2.

DALL-E 3 is currently available as a research preview and will be available to paid ChatGPT Plus subscribers and Enterprise customers in early October 2023. It will be made available to all users through OpenAI's labs website later this year.

Prior to this launch, DALL-E and ChatGPT were separate applications. However, DALL-E 3 has been built natively on top of ChatGPT, which lets people use the AI chatbot to create images from prompts ranging from a simple sentence to detailed paragraphs. One can also ask ChatGPT to make changes to the images with short commands, the company said.

This announcement comes on the heels of Google bolstering its chatbot Bard with deeper integration into several of its most popular consumer applications and services such as Gmail, Docs, YouTube, and Maps. Google also expanded Bard's visual search capabilities, including the ability to upload images with Google Lens and get Google Search images in responses to all languages supported by the chatbot.

In May 2023, Google said it will integrate Adobe’s family of creative generative AI models Adobe Firefly into Bard in the coming months, that will enable users to create high-quality images.

OpenAI said it has also taken steps to limit DALL-E 3's ability to generate violent, adult, or hateful content. For instance, the service will decline requests that ask for portrayal of public figures. It is also exploring ways to help people identify when an image was created with AI.

Earlier this week, OpenAI announced OpenAI Red Teaming Network, an initiative through which it is scouting for experts from various fields to evaluate their AI models and help inform its risk assessment and mitigation efforts at various stages of the model and product development lifecycle.

The company mentioned that DALL-E 3 is also designed to decline requests that ask for an image in the style of living artists while creators have the option of opting their images out from helping train OpenAI's future image generation models.