OpenAI Announces 'DALL·E 3', A Leap In Its Ability To Create AI-Generated Images


AI is about intelligence demonstrated by machines, as opposed to natural intelligence displayed by living things.

OpenAI is among the pioneers in the AI field, having previously created GPT-3, an AI capable of producing text rich with context, nuance and even humor, which is the successor of GPT-2.

Besides the highly-hyped ChatGPT, OpenAI also has what it calls the DALL·E, a portmanteau of the iconic artist Salvador Dali and the robot WALL-E from the computer-animated science fiction film of the same name.

The AI is meant to be the 'GPT' for images.

Its successor, DALL·E 2 is way more powerful.

This time, it's introduced 'DALL·E 3', which is even more powerful.

This the third version of its generative AI visual art platform allows users to use generative AI to create prompts and includes more safety options.

Before this, DALL·E and DALL·E 2 are able to convert text prompts to images. But even the newer DALL·E 2 can get things wrong, like ignoring specific wording. DALL·E 3 solves this, OpenAI researchers said, because it understands context much better.

By integrating DALL·E 3 with ChatGPT, users don't have to come up with their own detailed prompt.

Instead, they can just ask ChatGPT to come up with a prompt, and let the chatbot write out a paragraph for DALL·E 3 to follow. Since the ALL·E platforms work better with longer sentences, users can offload the creativity to the ChatGPT, and let DALL·E 3 do the rest.

"If you like a particular image, but it’s not quite right, you can ask ChatGPT to make tweaks with just a few words." explained OpenAI.

In an example, Aditya Ramesh, lead researcher and head of the DALL·E team, prompted ChatGPT to help him come up with a logo for a ramen restaurant in the mountains.

ChatGPT then wrote a longer prompt, and DALL·E 3 came up with several options to choose from.

"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," OpenAI said.

"Even with the same prompt, DALL·E 3 delivers significant improvements over DALL·E 2."

Just like before, OpenAI said that DALL·E 3 also focuses on safety.

The company said that it has taken steps to limit DALL·E 3’s ability to generate violent, adult, or hateful content. DALL·E 3 is also designed to decline requests that ask for a public figure by name.

DALL·E 3 is also designed to decline requests that ask for an image in the style of a living artist.

Starting DALL·E 3, creators can also opt their images out from training of OpenAI's future image generation models.

"We improved safety performance in risk areas like generation of public figures and harmful biases related to visual over/under-representation, in partnership with red teamers—domain experts who stress-test the model—to help inform our risk assessment and mitigation efforts in areas like propaganda and misinformation," OpenAI added.

Regardless, Sandhini Agarwal, a policy researcher at the company, said that she has "high confidence" in its safety measures, but clarified that the model continually improves and is not perfect.

The first DALL·E, released in January 2021, came before other text-to-image generative AI art platforms, like Midjourney, Google Imagen and more. But now, the competition is fierce.

Following the announcement of DALL·E 3, OpenAI said that the version shall be first be made available to ChatGPT Plus and ChatGPT Enterprise users in October.

Read: 'Red Teaming Network' Is Where OpenAI Invites Experts To Help It Improve Its AI