RT team
Google is actively developing artificial intelligence software aimed at mimicking human reasoning capabilities, drawing parallels with OpenAI’s o1 model, as the rivalry between these tech titans intensifies. Sources familiar with the developments at Alphabet Inc.’s Google, who prefer anonymity, reveal that the company is advancing in creating AI that excels in complex, multistep problem-solving tasks across various domains, including mathematics and coding.
Researchers at Google are exploring reasoning AI models, leveraging a technique called chain-of-thought prompting. This method, originally pioneered by Google, involves the software pausing briefly to process and evaluate multiple related prompts before delivering what it assesses as the most suitable response.
Despite the buzz, Google remains tight-lipped about its ventures in this area.
The competition with OpenAI has heated up, especially after the latter’s launch of ChatGPT. Google has responded by consolidating its research divisions into the Google DeepMind unit and strengthening the collaboration between its research and product teams. Nevertheless, Google’s approach to AI product launches is more measured, often pausing to address ethical considerations and the public’s trust expectations.
Notably, Google’s DeepMind expressed concerns about lagging behind following OpenAI’s o1 release. However, confidence within Google has been restored somewhat after showcasing some of their projects.
Google recently highlighted its AI’s capabilities in solving complex mathematical problems at the International Mathematical Olympiad and introduced Astra, an AI assistant at its developer conference, indicating its ongoing commitment to leading in AI technology. Demis Hassabis, CEO of Google DeepMind, emphasizes the importance of advanced mathematical reasoning in AI’s future.
