On Wednesday, Google unveiled its most ambitious venture to enter the rapidly expanding realm of generative artificial intelligence. This initiative introduces an AI model named Gemini, positioned to rival OpenAI’s GPT models, with the goal of enhancing various aspects, from Google’s consumer applications to Android smartphones.
The extent of Google’s aspirations is evident in the announcement, presenting Gemini as the company’s “largest and most capable AI model” and proclaiming the dawn of a “Gemini era,” envisioning widespread utilization across diverse settings, from major corporations to consumer gadgets like the Google Pixel 8 Pro.
In contrast to conventional AI models, which often specialize in handling specific user prompts like images or text exclusively, Gemini was designed as a “multimodal” system by Google. This implies that it can process inputs containing a variety of media types, integrating text, images, audio, video, and programming code.
“This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company,” said Google CEO Sundar Pichai in a blog post.
Google revealed on Wednesday that its exclusive AI chatbot, Bard, has undergone an upgrade incorporating the Gemini model. The company intends to integrate Gemini into widely used products such as Google’s search engine and Chrome web browser, both of which boast billions of users globally.
This announcement signifies an effort to regain control, as Google and other tech giants found themselves seemingly caught off guard a year ago by the sudden and immensely popular release of OpenAI’s ChatGPT. This event triggered a widespread industry rush to accelerate the development of generative AI tools and sparked a global discussion about the risks and benefits of AI.
Moreover, Google’s move reflects a strategic push to extend generative AI across its vast empire. The newly introduced Gemini 1.0 is available in three sizes: Nano, tailored for mobile devices and app developers; Pro, serving as the default model for a broad range of tasks and users; and Ultra, touted as Google’s most advanced AI model, currently undergoing safety testing.
Wednesday’s launch aimed to highlight Google’s advancements in cloud computing, a crucial tool for AI developers. The company revealed that it utilized a new generation of potent cloud-based processors to train Gemini, demonstrating a nearly threefold increase in speed compared to the previous version.
This technology, slated for availability to Google’s cloud customers, has the potential to significantly enhance the broader AI industry by making AI training more accessible. This move also strengthens Google’s standing as the third-largest player in the public cloud services market. However, the performance comparison between Google’s AI chips and those from leading manufacturers like Nvidia remains unclear.
During testing, Google reported that its Gemini model surpassed competing AI models in over two dozen commonly used benchmarks, assessing factors such as reading comprehension, mathematical prowess, and multistep reasoning skills.
“We do see it setting new kinds of frontiers across the board,” Eli Collins, vice president of product at Google DeepMind, told reporters on a conference call Tuesday.