Google's new AI can share its "thought process"

Google has released a new version of its artificial intelligence called Gemini 2.0 Flash Thinking, which can not only answer complex questions but also share its "thought process" (“reasoning”) while solving problems. According to The Verge, this reasoning- and analysis-focused model may rival advanced developments like OpenAI's o1.

A New Approach to AI

The standout feature of Gemini 2.0 is its ability to "think through" tasks before providing an answer. Jeff Dean, Chief Scientist at Google DeepMind, explained that the model is trained to break down complex instructions into smaller steps, enhancing the accuracy and efficiency of its solutions.

During a demonstration led by Dean, Gemini 2.0 applied this method to solve a physics problem. The model formed a sequence of intermediate steps, each bringing it closer to the final solution.

Although this is not fully akin to human reasoning, the approach allows the machine to structure tasks for better outcomes, notes The Verge.

Application Examples

One notable demonstration, presented by Google specialist Logan Kilpatrick, involved a problem combining textual and visual information. The model successfully interpreted both types of data, offering a solution that accounted for their interaction.

"Gemini 2.0 Flash Thinking is just the beginning of our journey to create machines capable of more logical reasoning," Kilpatrick stated.

Potential of the Technology

The Gemini 2.0 model represents the first step toward developing AI capable of solving tasks that require complex analysis. It has already received positive feedback for its innovative approach, which allows it not only to answer questions but also to showcase its thought process.

This could mark a significant breakthrough in artificial intelligence, especially for tasks involving data analysis from multiple sources.