On May 21st local time, American technology company Microsoft held its annual Build 2024 Developer Conference. Under the leadership of CEO Nadella, Microsoft's various product teams once again demonstrated astonishing execution power, presenting nearly 50 new products or feature updates at the press conference.
Nadella begins by mentioning that Microsoft has had two major dreams for decades: first, whether computers can understand humans instead of being forced to understand them, and second, whether computers can help humans reason, plan, and use all information more effectively in a world of increasing information. Nadella believes that the current wave of AI can provide Microsoft with answers to realize these two major dreams.
Subsequently, Nadella focused on Copilot. Nadella mentioned the Windows Copilot Library, which includes readily available local APIs, and stated that there are over 40 end-to-end AI models available out of the box, demonstrating how the new PC Copilot+PC, which fully integrates Copilot into personal computers (PCs), closely integrates with AI applications (apps).
On Monday, Microsoft introduced that the Windows Copilot runtime includes over 40 AI models. On Tuesday, Nadella likened Windows Copilot runtime to the heavyweight Windows API Wins2, as it made it easier for developers to use AI to build their own Windows apps.
Microsoft announces the release of Team Copilot and will release a preliminary preview version within the year
Microsoft announced the release of the latest Team Copilot, which is an extension of Copilot for Microsoft 365. It has changed from a personal AI assistant behind the scenes to an intelligent assistant in the team. Users will be able to call Copilot in collaborative applications such as Teams, Loop, Planner, etc.
According to the introduction, Team Copilot can act as a meeting host in meetings, managing agendas, tracking time, and taking notes. It can act as a collaborator in conversations, providing important information, tracking action items, and resolving unresolved issues. You can also act as a project manager to help ensure the smooth progress of each project and notify the team when they need their feedback. These preliminary experiences will be released in preview form later this year.
In addition, it is reported that Microsoft Copilot Studio has also introduced new proxy features, allowing developers to build Copilots that can actively respond to data and events, and tailor them to specific tasks and features. Copilots built with such new features can now utilize memory and knowledge to obtain contextual information, infer operations and inputs, learn based on user feedback, and seek help when encountering situations where they don't know how to handle them, thereby independently managing complex, long-term running business processes.
Azure AI Studio now supports the GPT-4o API and will be the first to use Nvidia Blackwell AI chips
Nadella introduced that the latest flagship model GPT-4o developed by OpeanAI is now available in Azure AI Studio and serves as an API. This multimodal model integrates text, image, and audio processing capabilities, bringing a brand new generative and conversational AI experience.
In addition, a new multimodal model, Phi-3-vision, developed by Microsoft in the Phi-3 Column AI Mini Language Model (SLM), has now been launched in Azure. The Phi-3 model is powerful, cost-effective, and optimized for personal devices. Phi-3-vision provides the ability to input images and text and receive text responses. For example, users can ask questions about charts or open-ended questions about specific images.
Developers can experiment with these cutting-edge models in Azure AI Playground and start building and customizing them using models in Azure AI Studio.
Subsequently, Nadella focused on Copilot. He mentioned the Windows Copilot Library, which includes readily available local APIs, and stated that there are over 40 end-to-end AI models available out of the box, demonstrating how the new PC Copilot+PC, which fully integrates Copilot into personal computers (PCs), closely integrates with AI applications (apps).
Nadella likened Windows Copilot runtime to the heavyweight Windows API Wins2 because it makes it easier for developers to use AI to build their own Windows apps.
Then Nadella mentioned AI infrastructure. He stated that Microsoft has "the most comprehensive artificial intelligence infrastructure in the world" and is expected to achieve 100% renewable energy power supply in its data centers by 2025. The number of countries where Microsoft Cloud Computing Azure's AI services are deployed has tripled.
He said that Microsoft will become one of the first cloud service providers to provide Nvidia's latest Blackwell AI chips, establishing a "deep partnership" with Nvidia.
Nadella stated that currently, over 50000 organizations are using Azure AI. With Azure AI, Microsoft offers the widest range of cutting-edge and open-source model choices, including Large Language Model (LLM) and Small Language Model (SLM).
Nadella mentioned at the developer conference that GPT-4o has been integrated with Azure AI, stating that OpenAI's GPT-4o demonstration last week was "very magical" and that OpenAI is Microsoft's "most strategic and important partner".
Meanwhile, Microsoft's AI SLM Phi-3 series has also released a new multimodal model, Phi-3-vision, which is now available for use in Azure. Microsoft introduced that the Phi-3 model is powerful, cost-effective, and optimized for personal devices. Phi-3-vision can input images and text, and receive text replies. For example, users can ask questions about charts or open-ended questions about specific images.
Microsoft announces the launch of Team Copilot, the latest extension of its Copilot series of generative AI technology.
Azure AI available text and image small model Phi-3-vision OPPO new mobile phone is currently testing Azure AI text language translation technology
The AI SLM Phi-3 series developed by Microsoft released a new multimodal small model Phi-3-vision on Tuesday, which can be used in Azure.
Microsoft introduced that the Phi-3 model is powerful, cost-effective, and optimized for personal devices. Phi-3-vision has 4.2 billion parameters and can input images and text, as well as receive text replies. For example, users can ask questions about charts or open-ended questions about specific images.
Microsoft also revealed that Chinese smartphone manufacturer OPPO is piloting Azure AI's voice to text, fast transcription, and Azure AI text to speech technologies on its new smartphones.
Team Copilot transforms a personal assistant into a team meeting host, chat collaborator, and project manager
Microsoft has launched multiple AI feature extensions for Copilot. A major expansion feature announced on Tuesday is Team Copilot.
As the latest extension of the Copilot series of generative AI technologies, Team Copilot has expanded Copilot for Microsoft 365 from a personal AI assistant behind the scenes to a new important member of the team. You will be able to call Copilot in collaboration tools such as Teams, Loop, Planner, etc.
Team Copilot is capable of serving as a meeting host, managing meeting agendas, tracking meeting time, and recording meeting points during meetings; You can also act as a collaborator in conversations, providing important information, tracking action items, and resolving unresolved issues. It can also serve as a project manager, helping to ensure the smooth progress of each project and promptly notifying the team for input. These initial experiences will be released in a preview version later this year.
Copilot Studio upgrade allows users to customize AI agents
Microsoft Copilot Studio has been upgraded with a new Agent Agent feature, allowing developers to build Copilots that actively respond to data and events based on specific tasks and features.
Based on these new features, Copilot learns context, infers operations, and inputs through memory and knowledge, learns based on user feedback, and seeks help when unsure how to handle them, thereby independently managing complex, long-term running business processes.
From IT equipment procurement to customer reception for sales and services, users can have Copilot built on Copilot Studio work for them. Before deployment, users can test their designed Copilot AI agent and then use it across multiple channels.
The first ARM Azure Cobalt based virtual machine preview version officially released Azure ND MI300X v5 virtual machine
Last November, Microsoft released its first self-developed AI chip Azure Maia 100 and its first self-developed cloud processor Azure Cobalt 100. On Tuesday, Microsoft announced the launch of a custom Cobalt 100 virtual machine (VM) preview version based on the Arm architecture, building upon Azure Cobalt 100. It is the first generation virtual machine to use Microsoft's new Cobalt processor, optimized for efficiency and performance when running general-purpose workloads and cloud native workloads.
Nadella introduced that compared to similar Azure virtual machines, the Cobalt 100 virtual machine based on Arm can improve performance by up to 40%, making it the most powerful computing product launched by Microsoft Azure.
Rani Borkar, the head of hardware systems and infrastructure at Microsoft Azure, also told the media that preliminary tests have shown that the performance of Azure Cobalt is up to 40% higher than the performance of commercial Arm servers currently used in its data centers. According to media reports, Microsoft is currently testing Cobalt CPUs on workloads such as Microsoft Teams and SQL Server, and plans to provide customers with virtual machines to handle various workloads next year.
Maia 100 is currently being tested on GPT 3.5 Turbo, which also supports ChatGPT, Bing AI workloads, and GitHub Copilot models. Similar to Cobalt CPUs, Microsoft is currently in the early stages of deploying the Maia 100 and has not disclosed its specific specifications and performance.
Microsoft is the first cloud service provider to use AMD's MI300X AI accelerator chip for customer AI training and AI inference. Microsoft announced on Tuesday that the Azure ND MI300X v5 virtual machine series has been officially released, optimized for high demand AI and high-performance computing (HPC) workloads such as Azure OpenAI Service.
Programming has entered the era of natural language, and programming novices can also develop software?
Nadella demonstrated GitHub Copilot Workshop on Tuesday, a software that helps novice programmers complete programming through chat. This will enable any beginner, even if they have little knowledge of programming languages, to develop their own software.
Nadella said, "Anyone can instantly turn ideas into code, and you can control every step of the process... This will make programming more enjoyable and interesting."
Microsoft plans to further refine this feature in the coming months to make it more usable.
AI real-time video translation function
Microsoft's Edge browser will soon launch a real-time video translation feature. Different from the current common video subtitle translation, this product has been able to provide real-time voice simultaneous interpreting translation.
Edge's AI real-time translation will initially support a range of mainstream video and news websites such as Bloomberg and Reuters, and only support bidirectional translation between English, Hindi, German, Russian, Italian, and Spanish. Microsoft also stated that it will support more websites and languages in the future.
As of the close on May 21st local time, Microsoft reported $429.04, an increase of 0.87%, with a market value of $3.2 trillion.
Daily Economic News Comprehensive Public Information
Disclaimer: The content and data in this article are for reference only and do not constitute investment advice. Please verify before use. Based on this operation, the risk is borne by oneself.