End of ChatGPT supremacy? Google is set to launch Gemini with significant upgrades.

Image credit : TheAIGRID Via YouTube Until now, OpenAI's prominent large language models have led the AI competition due to their early launch and the substantial data center infrastructure provided by Microsoft. Nevertheless, the current supremacy of ChatGPT might not endure indefinitely, given the continuous emergence of potent AI models on a monthly basis. Among these contenders, one particular model stands out as a potential formidable rival: Google. Based on a report from The Information, Google is poised to unveil its next-generation AI models as part of the Gemini project, possibly as soon as this autumn. Google intends to employ Gemini to enhance its existing AI chatbot Bard and to bolster enterprise applications like Google Docs and Slides. [caption id="attachment_190977" align="aligncenter" width="1920"] Image credit : Daffodil Software[/caption] What renders Gemini a robust adversary is the abundant resources that Google can harness, notably the extensive data that can be employed for training these AI models. Google holds access to a wealth of data, including YouTube videos, Google Books, a vast search index, and scholarly content from Google Scholar. Much of this data remains exclusive to Google, possibly granting it an advantage in constructing more intelligent models compared to its peers. A distinctive feature of Gemini is its status as the inaugural multi-modal model, capable of processing video, text, and images, a capability not shared by GPT-4. Furthermore, Google boasts a deep talent pool and years of expertise in the development and training of large language models. An announcement regarding the new models is anticipated in the upcoming autumn, likely accompanied by the introduction of a Gemini-powered chatbot or enhancements to the existing Bard chatbot. This has the potential to greatly impact Google's Cloud services, expected to be the primary avenue for corporate clients to access Gemini. [caption id="attachment_190979" align="aligncenter" width="607"] Image credit : Medium[/caption] The notion of Gemini was first introduced during Google's recent developer conference, where the company showcased several AI initiatives. Gemini draws from innovative training methodologies stemming from Alphago, an AI system recognized for defeating a professional human player in the intricate board game Go. This integration could empower Gemini's capacity to strategize and tackle complex problems. In April, Google amalgamated Google Brain, its in-house deep learning AI research division, with its subsidiary DeepMind, culminating in a single entity named Google DeepMind. This strategic move aimed to enhance efficiency by combining Google's extensive computational resources with DeepMind's research proficiency. Before the merger, both entities were pursuing their own responses to ChatGPT. DeepMind had Project Goodall, while Google was developing Bard, based on models from Google Brain. Despite initial competition between the teams, DeepMind abandoned Goodall to collaborate on Gemini. Also read : Google is reportedly working on AI life coach to provide helpful advice to people The collaboration between Google Brain and DeepMind could spell challenges for OpenAI and other competitors. The construction methodology of Gemini is also significant. Reports suggest that Gemini has made remarkable strides in its multi-modal capabilities, surpassing its predecessors. Its design prioritizes multimodality, enabling it to comprehend and generate different types of data, alongside being efficient in terms of tools and API integrations. This indicates that Gemini will excel not only in comprehending and producing conversational text but also in managing diverse inputs like text, images, and videos. There are indications that Gemini is being trained on double the number of tokens as GPT-4, which could potentially make it considerably more intelligent.