Thanks to their early introduction and Microsoft’s extensive data center infrastructure to support them, OpenAI’s huge language models have so far been outpacing competitors in the AI race. However, given the monthly development of new and potent AI models, ChatGPT’s dominance might not persist indefinitely. One of them, Google, stands a much better chance of competing seriously.
The Information claims that Google’s new Gemini project will introduce its next-generation artificial intelligence models as early as this autumn. The business chose Gemini to power its Bard AI chatbot and business tools including Google Docs and Slides.
Google owns a large portion of this data
The wealth of resources Google has at its disposal, particularly data that can be used to train these AI models, makes Gemini a fierce rival. YouTube videos, Google Books, a sizable search index, and academic content from Google Scholar are all accessible through Google. Since Google owns a large portion of this data, it may be able to develop models that are more intelligent than those of competitors. Gemini is reportedly the first multi-modal model that, in contrast to ChatGPT -4, can handle both text and video.
In addition, Google has a big talent pool and decades of experience in creating and refining sophisticated language models. The new models are anticipated to be unveiled in the autumn of 2019 along with the potential addition of a Gemini-powered chatbot or an upgrade to the Bard chatbot. The Google Cloud, which would probably be the primary method for business clients to access Gemini, could be significantly impacted by this.
At Google’s developer conference last month, where the company displayed a number of AI initiatives, Gemini was first alluded to. According to a story from June, Gemini employs fresh training methods from Alphago, an AI program that became the first to defeat a skilled human player in the challenging board game Go. Gemini might be able to plan and deal with issues as a result.
DeepMind and Google Brain are collaborating
Google combined DeepMind, a subsidiary firm, and Google Brain, a division dedicated to the study of deep learning artificial intelligence, in April to form Google DeepMind. Pichai’s unexpected decision was intended to increase productivity by integrating DeepMind’s research expertise with Google’s vast computer power.
Prior to that, each organization had its own reactions to ChatGPT. Both Google and DeepMind had Bard, which was based on Google Brain models. DeepMind dropped Goodall to work on Gemini despite some competition among the teams.
The fact that DeepMind and Google Brain are collaborating on this could be problematic for OpenAI and other competitors. It also concerns how Google is developing Gemini. According to reports, Gemini has significantly improved its multimodal capabilities compared to earlier models. Its architecture prioritizes multimodality, which allows it to analyze and comprehend many types of data. It also emphasises excellent tool and API integration efficiency.