Alphabet Inc. has revealed its groundbreaking artificial intelligence model, Gemini, designed to revolutionize information processing across diverse mediums like video, audio, and text.
With heightened capabilities in reasoning and nuanced comprehension, Gemini marks a significant leap forward compared to Google’s previous AI technologies, the tech giant announced on Wednesday.
“This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company,” stated Alphabet CEO Sundar Pichai in a blog post.
In a competitive response to OpenAI’s ChatGPT, Google intensified efforts to develop AI software that matches its rivals’ advancements. The company integrated a segment of Gemini into its AI assistant, Bard, aiming to launch the most advanced iteration of Gemini through Bard in the early months of the upcoming year.
Alphabet outlined plans to deploy three versions of Gemini, each tailored to leverage varying degrees of processing power. Ranging from data center-oriented high-power models to efficient iterations compatible with mobile devices, these versions aim to diversify accessibility.
According to DeepMind Vice President, Product Eli Collins, Gemini stands as the largest AI model crafted by Google DeepMind, boasting significantly enhanced cost-efficiency compared to its predecessors.
“It’s not just more capable, it’s also far more efficient,” Collins emphasized, highlighting the model’s improved efficacy alongside demanding computational requirements during the training phase.
Moreover, Alphabet announced a fresh iteration of its custom-built AI chips, termed tensor processing units (TPUs). The Cloud TPU v5p, structured in pods comprising 8,960 chips, targets accelerated training for large-scale AI models. The new TPUs purportedly accelerate the training of expansive language models by nearly threefold compared to prior generations.
As of Wednesday, the enhanced chips are available for developers in a “preview” phase, showcasing Alphabet’s commitment to advancing AI infrastructure and accessibility.
Alphabet’s strides with Gemini and the upgraded TPUs underscore the company’s commitment to pushing the boundaries of AI development while addressing the critical aspects of efficiency and accessibility in this technology realm.
