Bloomberg reported yesterday (October 2) that,GoogleCompany is developing human-like reasoning AI, hoping to catch up in terms of progress OpenAI The company's o1 model.
OpenAI recently introduced o1-preview and o1-mini models to ChatGPT, which can simulate human thinking patterns, realize universal reasoning, and allow AI to "think".
Citing Bloomberg, Google Inc. is also now aggressively pushing to land a generalized reasoning AI model that is better at handling math and programming, though it consumes more processing power and energy.
Like OpenAI, Google is trying to get closer to human reasoning using a technique called "chain-of-mind hinting," people familiar with the matter said. In this technique, pioneered by Google, software pauses for a few seconds and then responds to written prompts, while in the background, invisible to the user, it considers a number of related prompts and then summarizes the answer that seems best.
The new model utilizes step-by-step reasoning cueing techniques and is able to consider multiple responses and debate itself to find the correct solution.
Bloomberg said that Google is not too far behind in the general reasoning AI model, there are already more adequate technical reserves, the article did not mention the release time, only revealed that Google's development of the AI model will be in the "next few months" have new progress.