Alphabet (NASDAQ:GOOGL) unveiled its highly anticipated artificial intelligence model, Gemini, marking a significant leap in AI capabilities. Gemini boasts advanced processing capabilities, enabling it to analyze diverse data forms, including video, audio, and text, with enhanced reasoning and nuanced understanding compared to Google’s previous AI technology.
In a blog post, Alphabet CEO Sundar Pichai expressed that Gemini represents one of the company’s most substantial science and engineering endeavors. This release comes in response to the competition posed by OpenAI’s ChatGPT, prompting Google to strive for AI software that rivals its Microsoft-backed counterpart.
The Gemini model has already been integrated into Google’s AI assistant Bard, and Alphabet plans to roll out the most advanced version of Gemini through Bard early next year. The company has crafted three versions of Gemini, each optimized for different processing power requirements, ranging from data center deployment to efficient operation on mobile devices.
Notably, Gemini stands as the largest AI model developed by Google DeepMind AI unit, yet it is reported to be “significantly” more cost-effective than previous larger models. According to DeepMind Vice President, Product Eli Collins, Gemini is not only more capable but also more efficient, requiring substantial computing power for training but demonstrating improvements in the process.
In addition to Gemini, Alphabet announced a new generation of its custom AI chips, the Cloud TPU v5p. Designed for training large AI models, these chips are arranged in pods of 8,960 and are capable of accelerating language model training nearly three times faster than prior generations. As of Wednesday, the new chips are available for developers in a “preview” stage, reinforcing Alphabet’s commitment to advancing AI technology.
Featured Image: Megapixl