Google introduced an array of fresh artificial intelligence (AI) technology and collaborations on Tuesday, with a focus on expanding the adoption of this rapidly growing technology among large enterprises, Reuters reports.
During its Google Next conference in San Francisco, the tech giant unveiled a series of announcements, including new clients for its cloud software such as General Motors and Estee Lauder Companies.
The subsidiary of Alphabet publicly revealed an updated version of its proprietary AI chips, introduced a comprehensive tool for enterprises to watermark and identify AI-generated images, and showcased security tools and enhancements for its office suite. These announcements are part of Google's recent efforts to showcase its AI strategy, particularly after Microsoft surprised the company with an ambitious AI plan it has been implementing since the previous year.
However, Thomas Kurian, Chief of Google Cloud, emphasized that the adoption pace for big business clients needs to be deliberate and distinct. He advised enterprises to proceed methodically and strategically in their approach to AI adoption. He stated, "There's been this sort of FOMO of, 'I need to be in generative AI for generative AI's sake.'" FOMO, or "fear of missing out," is a prevalent sentiment in the AI field.
To bolster its enterprise cloud service, Google integrated 20 new AI models into its repertoire, increasing the total to 100. This expanded AI infrastructure includes partnerships to grant Google Cloud customers access to AI model LLaMa 2 from Meta Platforms, as well as access to Claude 2 from the startup Anthropic.
Furthermore, Google introduced updated versions of its foundational AI infrastructure, enhancing performance and adding new features. An example is the new iteration of its text model PaLM, which enables users to input larger amounts of text, making it more feasible to process lengthy documents like legal briefs and books. The company also unveiled SynthID, a tool that facilitates watermarking of AI-generated images, ensuring their integrity even after alteration.
AI enhancements were also extended to Google's suite of office software and security tools. Notably, the company presented an AI-powered solution capable of transitioning databases from Oracle to an open-source version, a task known for its complexity.
In anticipation of the launch of its full-fledged fifth-generation tensor processing unit (TPU), Google introduced an access version tailored for genAI and large language models. Known as TPU v5e, this chip is designed to efficiently train and serve large models. It's distinct from the yet-to-be-released flagship fifth AI generation chip. Google has grouped TPU v5e chips into 256-unit batches, referred to as a "supercomputer." Cloud users can connect multiple pods together to address more intricate computing challenges.
Follow Daryo's official Instagram and Twitter pages to keep current on world news.
Comments (0)