"Mouse pointers should be used as AI input," says DeepMind: Are we entering an era where saying 'make this into that' will work with AI?

Google DeepMind is exploring a new AI input model that combines the mouse pointer, on-screen context, and short voice instructions.

Users could point at items in webpages, documents, images, or maps and ask AI to summarize, compare, edit, or navigate.

The concept is being demonstrated with Gemini and may appear in products like Gemini in Chrome and a future Googlebook feature called Magic Pointer.

The goal is to make AI help feel more natural and faster by reducing the need for long written prompts.