Google shows off AI model that thinks

Google has unveiled a new artificial intelligence model that is capable of answering complex questions while also providing an overview of its “thoughts.” The model, called Gemini 2.0 Flash Thinking, is still in the testing phase and is likely to compete with a similar model from OpenAI.

In a post on X, Google DeepMind chief scientist Jeff Dean says the model is “trained to use thought to enhance its reasoning,” and it also benefits from the speed of the faster Gemini Flash 2.0 model. The demo Dean shared shows how Gemini 2.0 Flash Thinking answers a physics problem by “thinking” through a series of steps before offering a solution.

This is not necessarily “thinking” as humans do, but rather means that the machine divides instructions into smaller tasks that can yield better results.

Want to see Gemini 2.0 Flash Thinking in action? Check out this demo where the model solves a physics problem and explains its reasoning. pic.twitter.com/Nl0hYj7ZFS

— Jeff Dean (@JeffDean) December 19, 2024

Another example, posted by Logan Kilpatrick, a product manager at Google, shows a model thinking about a problem, incorporating both visual and textual elements. “This is just the first step in our thinking journey,” says Kilpatrick. You can try out Gemini 2.0 Flash Thinking in Google’s AI Studio.

There have been some notable updates in the AI space recently, with Google earlier this month unveiling an upgraded version of Gemini 2.0, part of the company's efforts in agent-based AI, and OpenAI making its o1 language inference model available to ChatGPT subscribers.