Microsoft’s Bing AI chatbot gets support for image creation

Microsoft’s Bing AI chatbot can create images based on text input from the user. Users can provide a description of an image, along with additional details. Bing uses a version of OpenAI’s Dall-E model for the images.

The image generator is called Bing Image Creator and is in the Bing chat and in the Edge browser. For now, the Image Creator only works when users start a conversation in Creative mode, not in Balanced or Precise modes. Users can write a description of an image, provide additional context such as a location or an activity, and choose an art movement. To draw a picture in chat, users must say something like “draw a picture” or “take a picture.” Bing Image Creator can then create a newsletter or provide inspiration for redesigning a room, Microsoft gives as examples.

The Image Creator uses an “advanced version” of OpenAI’s Dall-E model, “with additional protections.” For example, Microsoft says that the Bing Image Creator should create as few harmful or unsafe images as possible. If the generator thinks that the user wants to do this, a warning will appear on the screen. Microsoft also wants to indicate with a Bing icon in the bottom left corner that an image was created with the Image Creator, to warn others.

Microsoft says it will only release the Image Creator to ‘certain’ preview users for the time being, with a view to bringing it to other users later. The Bing Image Creator works on both desktop and mobile. In the future, the image maker should also work in Balanced and Precise modes, and with more languages.

Microsoft also says it will bring Stories and Knowledge Cards 2.0 to Bing search. Stories are images and short videos that should appear in searches, while Knowledge Cards 2.0 are infographic-style cards that should provide “fun facts and information” in searches. The 2.0 version uses AI to show ‘interactive, dynamic’ content such as graphs and timelines.

The Stories (left) and Knowledge Cards 2.0