OpenAI is about to launch the third version of DALL-E, its artificial intelligence system capable of transforming text into images. This new iteration promises to surpass the capabilities of its predecessors. But most of all, DALL-E 3 allows integration with ChatGPT, also developed by OpenAI, to generate more precise prompts and requests for the image generator.
DALL-E integrates ChatGPT, the two OpenAI AIs work together
According to what was stated to Axios and demonstrated to The Verge, DALL-E 3 better understands users’ textual requests when generating images. This is because, in case our requests were unclear to the OpenAI algorithm, DALL-E will ask its “brother” bot for help.
In case a user finds it difficult to express what they want in order to make the most of the image generator’s capabilities, ChatGPT can step in to help you formulate a complete message on this matter. In practice, if you lack the vocabulary and precision to get the right prompt for the image you want, ChatGPT will rephrase your sentence so that DALL-E understands exactly what you mean.
But novelty also helps you find more iterations of the same idea. During the demonstration on The Verge, DALL-E delivered four convincing results for a request to create a logo for a ramen restaurant located in the mountains. To do this, he leveraged ChatGPT to change the prompt slightly and find new possibilities.
AI will not imitate living artists
The integration of ChatGPT is just one of the new features of DALL-E 3. In fact, the new version of the bot allows you to better address the challenges related to creating complex elements, such as depictions of hands and texts within images. The company has also placed particular emphasis on security measures, training DALL-E 3 to ignore certain words in messages that could lead to generation of explicit or hateful images.
OpenAI has taken action to prevent potential legal problems, prohibiting the new version of DALL-E from imitate the work of living artists. The artists themselves will have the opportunity to send their works via a dedicated form on the site e request their removal if they are replicated or used improperly.
The official launch of DALL-E 3 is scheduled for next month for ChatGPT Plus subscribers and for corporate customers. Subsequently, the new image generator will be made available for i OpenAI research laboratories and for API customers in the autumn of this year.
At the moment, the company has not yet announced the release date for the general public, so at least for the moment it will remain a version for those who pay a subscription or do research (come GPT-4 per ChatGPT).
What however seems like an important change of pace is the integration of two different OpenAI AIs to improve the results. Creating, in fact, an ecosystem of artificial intelligence which could attract new subscribers to the service.