Google to take on OpenAI’s o1-preview? Search giant working on ‘AI reasoning model’

Reportedly, several teams at Google are working on an AI model that has the ability to solve complex multi-step problems.

Google is reportedly developing an AI model that can think before it responds. (Image: DALLE)

The AI race between OpenAI and Google is about to get more intense. Weeks after OpenAI launched its new series of reasoning models called the OpenAI o1-preview, Google is now reportedly working on its own AI models with advanced reasoning capabilities.

Multiple teams at Google are working on an AI model that has the ability to solve complex, multi-step problems, according to Bloomberg. The AI model employs chain-of-thought prompting, a method created by Google to take on complex problems in maths and programming. The AI model reportedly ‘thinks’ before responding.

However, Google is taking a more cautious approach with this model than OpenAI did with the o1. The search giant has already released its math-focused reasoning models, AlphaProof and AlphaGeometry 2.

Google and OpenAI have been eyeing the top spot in the AI arms race ever since the launch of ChatGPT. After ChatGPT became the fastest app to attain the most users, Google took various measures to stay afloat.

Also Read | OpenAI DevDay 2024: Four new features to make AI more affordable and accessible

In July, the tech giant introduced AlphaProof, a model that specialises in maths reasoning, and AlphaGeometry 2, a model focussed on geometry. Similarly, in May, the company provided a glimpse of its AI assistant Astra, which is an AI model that can use a phone’s camera to see its surroundings and answer questions. All these features, Google had said, would be coming to Gemini by the end of this year.

Lately, giving their AI models human-like reasoning and agent-like capabilities seems to be what most big AI firms are betting on. The release of OpenAI o1-preview has in a way signalled a new phase in the LLM race. The bigger question is will OpenAI further its competitive edge and make it difficult for its peers to come out with top-tier models.

Tags:
artificial intelligence