OpenAI has integrated Dall-E 3, its latest image generator, into ChatGPT. The tool is currently available in beta for subscribers to ChatGPT Plus, OpenAI's $20-a-month service. With this image generator at your disposal, you can easily instruct the chatbot in plain language to craft a set of four unique images.
However, as these powerful image generators become accessible to the public, concerns regarding legal and ethical implications have come to the forefront. This integration also raises red flags among security experts who worry about AI image generators contributing to the spread of disinformation.
For those eager to explore GPT Dall-E 3 without a subscription, Microsoft's Bing Image Creator offers a version to experiment with. In the initial days of this integration, some users created controversial imagery using Bing, highlighting the need for safeguards. Subsequently, Microsoft has implemented stricter controls around the AI image generator.
But if you're a ChatGPT Plus subscriber, enabling the chatbot's Dall-E 3 features is straightforward. You need to log in to OpenAI's website or use the ChatGPT mobile app, available on both Apple and Android platforms. After accessing ChatGPT, the next step is to click on the GPT-4 tab located at the top of the screen. In the dropdown menu that appears, select Dall-E 3 (Beta).
Be mindful of a usage cap governing interactions with the GPT-4 version of ChatGPT. These prompts demand substantial computational power to process. The official limit stands at 50 prompts every three hours. Should you reach this limit, the chatbot will display an error message, specifying the duration of the waiting period before you regain access.
Beyond the rate limit, expect an approximate 30-second wait for the images to materialize. If any of the generated creations contravene OpenAI's guidelines, you may receive compliant images or even a message refusing the request.
For those who have previously explored AI image generators like Midjourney or Dall-E 2, the integration of Dall-E 3 introduces a significant difference. It allows you to witness ChatGPT's role as an intermediary, creating multiple prompts for Dall-E 3 to complete.
These prompts, crafted by ChatGPT, encompass a wide range, from long sentences to complete paragraphs, each supplying diverse details for GPT Dall-E 3. If individuals are part of the image, the chatbot often explicitly mentions gender and race when describing the subjects.
If the initial results generated by the chatbot do not align with your vision, you can ask for adjustments in various aspects such as color schemes or overall aesthetics. For instance, if you're particularly fond of the third image produced by Dall-E 3, you can request more images that resemble it by clicking the download button in the top left corner.
One pressing question remains: Has anything been done to protect artists in this new update? In reality, not much. While the chatbot refrains from creating images that mimic contemporary artists, there are still workarounds. For instance, when asked to design a coffee mug in Keith Haring's style, ChatGPT declined the initial prompt but offered an alternative, stating, "I can create a design inspired by the general characteristics of his art, such as bold lines, vibrant colors, and simplistic figures. Would you like me to proceed with that?" The results, however, were often less than satisfying, appearing messy and mediocre.
Despite the advancements in image quality with GPT Dall-E 3, underlying issues with image generators persist. Users should expect to encounter peculiar distortions and eerie faces in the images it produces. Some issues may be humorous, such as the chatbot's struggle to label baking ingredients accurately, while others are more concerning.
If you are a person with a keen interest in the dynamic field of AI, follow Atlasiko news and learn more!
