Google is taking its rivalry with OpenAI to another level

Google headquarters on August 13, 2024 in Mountain View, California. - Photo: Justin Sullivan (Getty Images)
Google headquarters on August 13, 2024 in Mountain View, California. - Photo: Justin Sullivan (Getty Images)

Google (GOOGL) reportedly wants to bring its artificial intelligence rivalry with OpenAI to the next level of AI models: those that can “reason.”

Teams at Google have made progress in recent months on software for AI models that resembles humanlike reasoning abilities, Bloomberg reported, citing unnamed people familiar with the matter. Google has long focused on “reasoning” capabilities in large language models (LLMs), including in its work on chain-of-thought prompting.

With this technique, which people told Bloomberg that Google is using, LLMs can solve multistep problems using “a series of intermediate reasoning steps,” similar to the way a human would. Models with this software take a longer time to respond to inquiries because they consider similar prompts that are summarized into a response. The chain-of-thought technique gives models the ability to solve more complex math- and computer-programming-related inquiries.

OpenAI is also using chain-of-thought prompting for its new o1 model. After the model, which was internally called Strawberry, was released in September, some employees in Google’s DeepMind unit were worried the company had fallen behind, a person told Bloomberg. However, they added that employees are no longer concerned now that Google has unveiled more competitors to OpenAI’s products.

Unlike in the current version of ChatGPT, OpenAI’s new model doesn’t have some of its “useful” features yet, such as browsing the web and file and image uploads. The o1 series models are “designed to spend more time thinking before they respond,” the company said. The models, the first of which are available in preview in ChatGPT and through the company’s API, can “reason” through more complex tasks and problems in science, coding, and math than earlier OpenAI models can.

Google is also working on improving its Gemini chatbot’s “reasoning” abilities. In July, Google made its fastest, most cost-efficient model, 1.5 Flash, available in the unpaid version of Gemini. With 1.5 Flash, Gemini will have “quicker and more helpful responses,” Google said, adding that users will notice improvements to Gemini’s reasoning and image processing abilities.

For the latest news, Facebook, Twitter and Instagram.