Does the New DALL-E 2 Program Indicate More Creativity in AI?
The DALL-E 2 program is an image generator of fantastic images from text-based descriptions and whatnot.
OpenAI, a San Francisco-based company best known for its large GPT-3 natural language model, announced that it will release a second version of its text-to-image AI model. Like its predecessor, the new DALL-E 2 program is a neural network that creates images based on user-entered natural language phrases. However, while the original image of DALLE was low resolution and conceptually simple, the image generated by DALL-E 2 is five times more realistic and accurate using creativity in AI. Also, the second DALL-E is actually a smaller neural network. (OpenAI refused to specify the dimension of DALL-E 2 program in the parameter.)
DALL-E 2 program is also a multimodal neural network, which means it can process both natural language and visual images. For example, you can display two different images in your model and ask your model to create an image that combines different aspects of the source image in different ways and the creativity that the system seems to show in doing so is, well, a little embarrassing. DALL-E 2 program was given two photos one looked like street art and the other looked like Art Deco, to quickly create a set of about 20 images placed on the grid, each different from adjacent images. The system combined different visual aspects of the source image in different ways. Some seemed to be able to fully express the dominant style of one source image while suppressing the other. In summary, the new image had a different design language than the source image.
OpenAI engineers struggled to explain the steps to take to prevent the model from creating unwanted or harmful images. OpenAI researcher states that nude, violent, and bloody images have been removed from the training dataset. Without it, it is “very unlikely” that the DALL-E 2 program will inadvertently produce such a thing. OpenAI people also monitor images created by users with the DALL-E 2 program such as adult, violent and political content is not allowed on the platform. OpenAI says it plans to regularly roll out get right of entry to the brand new version to businesses of “trusted” users. Eventually, the tea is hoping to get right of entry to the DALL-E 2 program through an API [application programming interface]. Developers will then be capable of constructing their very own apps on the pinnacle of the AI version.
Looking at the actual application of the model, DALL-E 2 program could help graphic designers who can use the tools to open new creative avenues. And finally, developers who access DALL-E 2 program via the API may find new and novel uses for this technology. DALL-E 2 program can be an important tool. Creating a language is natural for humans, but creating images is not that easy. Vision and language are both key parts of human intelligence; building models like DALL-E 2 program connects these two domains. It’s a very important step for us as we try to teach machines to perceive the world the way humans do, and then eventually develop general intelligence.