Competition in artificial intelligence development is getting fiercer. OpenAI and Google are two of the leading names in the AI segment. OpenAI got off to a head start, with Google trying to catch up with Gemini and other developments. In line with this, new reports claim that Google is working on boosting the reasoning capabilities of its AI model to match ChatGPT’s o1.
Google seeks human-level AI reasoning model to compete against ChatGPT’s o1
OpenAI announced o1 this month as a new model integrated into the ChatGPT service. The model is not a successor to previous models like GPT-4 or GPT-4o, but a different approach. The o1 model is designed to offer human-like levels of reasoning, taking more time to “think” in order to deliver a more detailed output. o1 seems especially useful for complex requirements or problems related to areas like math and coding. OpenAI was working on o1 under the code name “Strawberry.”
Currently, Google does not have anything that can compete against OpenAI’s o1 AI model. However, according to Bloomberg, the Mountain View giant is already working on an advanced reasoning AI model to integrate into its services. Like OpenAI, the new model would not be the company’s next big leap. It would be more like an improvement to its existing AI models. Google does not want its Gemini AI reasoning model to lag far behind the competition.
Details about the new model are still scarce
The report reveals that Google is using a technique called chain-of-thought prompting. The company pioneered this technique, which consists of AI taking into account not only the user’s main prompt but also a set of related prompts to offer a complex response in the form of a summary. At present, there is no info on the availability or even the name of Google’s new model. Hopefully more details will emerge in the coming weeks.
