AI research lab OpenAI has just unveiled DALL-E 3, the third iteration of its AI image generator with even more potential for interpreting nuance and detail than DALL-E 2.
Currently in research preview, DALL-E 3 also integrates directly with ChatGPT. This allows the chatbot to do the heavy lifting by creating prompts, opening its use up to a less tech-savvy audience.
OpenAI's soon-to-be-launched image creator also boasts enhanced safety measures to reduce algorithmic bias and the generation of harmful content. But whether DALL-E 3 is doing enough to assuage the concerns of independent creators remains yet to be seen.
OpenAI Unveils DALL-E 3: its User-Friendly, Advanced AI Image Creator
ChatGPT creators OpenAI recently launched its first official preview of DALL-E 3 – its most sophisticated and easy-to-use image generation program to date.
According to its new release page on OpenAI's website, the text-to-image generator understands “significantly more nuance and detail” than previous systems, making it easier than ever for users to translate ideas into accurate images.
Unlike all previous iterations of DALL-E, this new model integrates with the company's chatbot, ChatGPT, too. This allows users to use ChatGPT to create tailored and detailed prompts, making it easier for users with less technical know-how to bring their ideas to life.
🔎 Want to browse the web privately? 🌎 Or appear as if you're in another country?
Get a huge 86% off Surfshark with this Tech.co Black Friday offer.
“Modern text-to-image systems have a tendency to ignore words or descriptions, forcing users to learn prompt engineering. DALL·E 3 represents a leap forward in our ability to generate images that exactly adhere to the text you provide.” – ChatGPT website
OpenAI has also taken additional steps to limit DALL-E 3's ability to generate violent, adult, or hateful content, and improved safety performance in risk areas like harmful biases around over and under-representation.
Since OpenAI launched its seminal text-to-image generator DALL-E in January 2021, the market has become rife with a number of high-quality copycats. A number of platforms like Midjourney and Stable Diffusion have gained more notoriety than DALL-E due to their enhanced image quality and customization options.
However, OpenAI believes this product update could cement DALLE-3 in prime position, due to its powerful ability to generate output that adheres exactly to the text the user provides.
Is DALL-E 3 More Friendly to Creatives?
Since DALL-E's initial launch in 2019, its training process, and potential to infringe on copywritten work, have been the source of much controversy.
Similarly to ChatGPT, DALL-E 2 was trained on millions of images scraped off the web without necessarily gaining the consent of their owners. While this unlocked an abundance of rich source material, it also sparked a debate over the ownership of AI-generated images, as well as a number of lawsuits brought about by disgruntled artists and image creators.
To avoid these ethical quandaries, OpenAI is doing things a little differently with DALL-E 3. The text-to-image generator will reject requests that ask them for an image created in the style of living artists. The tool is also allowing creators to opt their images out from training their future image generation model.
OpenAI efforts clearly mark a step in the right direction, especially for creatives looking to remove their art from DALL-E's training set permanently. However, for those who believe their art has already been stolen through OpenAI's invasive scraping practices, these amendments do little to quell their concerns.
When is OpenAI's DALL-E 3 Available?
DALL-E 3 is currently limited to research preview, which means that its use is currently not available to the public.
However, DALL-E's newest version will be released to ChatGPT Plus and ChatGPT Enterprise users throughout October, shortly followed by research labs and its API service later in the fall.
OpenAI has not commented on when a free version of DALL-E 3 will be released.