OpenAI has recently made its newest image generating AI model, DALL-E 3, accessible to users of its ChatGPT Plus subscription service and ChatGPT for Enterprise. Building on its predecessor, DALL-E 3 goes beyond mere visual appeal and delivers images with enhanced clarity and detail.
DALL-E 3 is capable of rendering intricate elements like text, hands, and faces, producing visually striking outputs. It excels at responding to detailed prompts and supports both landscape and portrait aspect ratios. Moreover, it outshines its competitors, such as Adobe Firefly 2 and Midjourney, by offering users the ability to generate text and typography directly embedded in images. This functionality proves advantageous for marketing, branding, and business-related visual content like promotional imagery and sales materials.
OpenAI has provided several real-world examples to showcase DALL-E 3’s potential applications within the ChatGPT Plus environment. From generating art for school projects to creating corporate logos, the possibilities are vast.
Conversational Interaction and Image Manipulation
In VentureBeat’s tests of the DALL-E 3 ChatGPT Plus integration, one standout feature is the AI’s conversational capabilities. Users can engage in a conversation with the AI, requesting modifications to images, such as moving elements or making changes. This functionality eliminates the need to generate entirely new images or manually edit sections, providing a seamless and efficient user experience.
OpenAI’s dedication to responsible AI development is evident in their recent release of a research paper detailing the development of DALL-E 3. Additionally, they are working on an image classifier that can accurately determine if an image was generated by DALL-E 3 with an impressive 95-99% accuracy rate. This classifier serves as a valuable tool in combatting AI-generated disinformation and propaganda, which have seen a surge amid the Israeli-Hamas conflict.
“We’re researching and evaluating an initial version of a provenance classifier – a new internal tool that can help us identify whether or not an image was generated by DALL-E 3. Despite these strong results on internal testing, the classifier can only tell us that an image was likely generated by DALL-E, and does not yet enable us to make definitive conclusions.”
OpenAI’s responsibility-driven approach becomes evident with the development of this provenance classifier. It showcases their commitment to understanding and mitigating potential negative impacts of AI-generated content on society. This initiative follows the release and subsequent withdrawal of a classifier for AI-generated text, which was deemed inaccurate and mistakenly labeled human-generated text as AI-generated, particularly affecting non-native English writers.
Although no release date has been announced for the AI image classifier, OpenAI’s proactive efforts demonstrate their dedication to accountability and technological advancement.
