Microsoft Highlights End of Year AI Gains and Acquisitions

This week, Satya Nadella, Chairman and CEO at Microsoft, posted a video on his LinkedIn account, reflecting on recent successes in bringing the firm’s AI vision to enterprise clients – with a mention of Microsoft’s recently updated Phi service and CoHere acquisition that optimize AI models for businesses.

Nadella said:

There is no doubt, 2023 was the year of AI. We are no longer just talking about AI innovation in the abstract. We are seeing real product making, deployment, and productivity gains from breakthroughs in silicon to the creation of small language models like Orca and Phi, two models as a service which is making it easier to bring CoHere, Llama, Mistral to more apps – to the latest and the best from Open AI and of course CoPilot.

The Microsoft CEO noted how the firm’s current lineup of AI services and partnerships can drive optimization, accessibility, and sustainability in the workplace.

Nadella also said AI innovation will “only be useful if we build and use it responsibly, ensuring its benefits empower each of us in our careers in our communities, and in our countries while helping to address some of the world’s most pressing challenges.”

Nadella mentioned several services in his video. Alongside the well-noted OpenAI/CoPilot collaboration, Nadella mentioned accompanying Microsoft enterprise-grade AI services, including Orca, Phi, CoHere, Llama, and Mistral, with each service representing recent gains for Microsoft.

While Orca is a long-standing service, most other mentioned AI services are new or updated. In recent months, Microsoft recently added many of these AI Model-as-a-Service (MaaS) products to its Azure platform. Microsoft is working on each service to help optimize its AI offerings for the enterprise client base Nadella mentioned in his video, such as government, finance, healthcare, and education.

Microsoft Integrates Command by CoHere

Satya Nadella’s mention of CoHere comes following Microsoft’s November acquisition of CoHere’s enterprise AI MaaS platform, Command, with the firm adding it to the Azure umbrella.

The merger followed initial announcements of the acquisition made by Microsoft at Ignite earlier this year. The move enables Microsoft to launch the Command service to an international audience with Azure Cloud support included.

Jaron Waldman, Cohere’s Chief Product Officer, said at the time:

Bringing Command as a Service to Microsoft Azure is an important step forward in our goal of meeting our customers where their data lives. With so many Azure customers working to scale their GenAI prototypes toward production there’s an immediate need for highly efficient, accurate model options. The ability to access Cohere’s enterprise models seamlessly on Azure makes it easy for them to scale AI affordably and transform their businesses.

The acquisition further enables Microsoft to create targeted AI solutions to match a range of enterprise use cases.

Microsoft Optimises Enterprise AI Models

Alongside the gain of CoHere’s AI MaaS service, Microsoft also recently unveiled its most recent version of Phi earlier this week, a small language AI model.

Phi-2 is an in-house text-to-text AI service, ready on Azure, that Microsoft says has “remarkable performance” when on several small computing devices, potentially increasing accessibility to high-quality enterprise-grade AI services.

In addition to Phi-2, Microsoft introduced Mistral in November, a MaaS that supports large models and extensive response times via “grouped query attention and sliding window attention.”

Moreover, in July, Microsoft collaborated with Meta to expand the Llama 2 AI language models (LLMs) on Azure and Windows platforms.

Llama 2 is a project that enables developers and organizations to create generative AI-powered tools and experiences- now, Microsoft Azure customers can customize and deploy Llama 2 models on Azure.

Meta and Microsoft have a long history of working together on AI development. This includes collaborating to integrate ONNX Runtime with PyTorch to create a more refined developer experience for PyTorch on Azure, as well as Meta’s decision to choose Azure as their strategic cloud provider.



from UC Today https://ift.tt/UMEp28O

Post a Comment

0 Comments