Kakao Brain, the artificial intelligence technology research and development subsidiary of Kakao, said Wednesday it has launched its second AI model that self-generates images, allowing the company to quickly expand its presence in Korea’s mammoth AI space.

The AI model, dubbed “minDALL-E,” creates images the users wish to see, according to the company. For instance, when users type in requests such as “draw a picture in a Salvador Dali style” or “create an image using a full moon and the Eiffel Tower,” the AI model presents such images.

Unlike a search engine that simply finds already existing images on the web, the AI model is capable of understanding the context of the users’ requests and creating a whole new image, the company said. Prior to the launch, the AI model has undergone a machine learning process with 14 million sets of text and images.

“Since minDALL-E mimics the art styles of popular artists, rather than replicating the artwork itself, there is no problem of copyright infringement,” a Kakao Brain official said.

With the AI model, users can easily generate illustrations for storytelling content or to make educational materials.

“If we join hands with KakaoTalk, users might even be able to self-create their own emoticons,” the official said.

Kakao Brain has recently upgraded its first Korean-language AI model KoGPT as well. Using the deep learning supercomputer to carry out calculations with larger data sets, it has enhanced the model’s accuracy to better understand Korean, the official said.

On Friday, Kakao Brain will hold an online miniconference to share their insights on KoGPT and minDALL-E.

“We are researching the components required for an AI to think and speak like a human being. Also, we ask ourselves how an AI that has memory and deduction skills could communicate like us,” said Kim Il-du, CEO of Kakao Brain.

“Starting with the two AI models, we plan to unveil other supergiant AI models as open source to further contribute to the development of the country’s IT industry and the next generation of deep learning,” Kim added.