Microsoft expands Copilot and plug-in ecosystem for developers

Microsoft recently announced a major expansion of the AI ​​plug-in ecosystem and will adopt the same open plug-in standard as OpenAI ChatGPT.

At the same time, Microsoft also launched a new series of Copilot tools. These applications use generative AI to help users complete cognitive tasks, such as writing sales pitches, making Take notes or generate images.

Following these announcements at the Microsoft Build 2023 conference, Microsoft has also made a wave of AI-centric updates to the Microsoft Azure cloud platform, aiming to improve developer productivity and help enterprises harness the power of AI. achieve your own goals.

Expanded Plug-in Ecosystem

Two years ago, Microsoft introduced the concept of "AI-powered Copilot" when it launched GitHub Copilot, and Copilot is a help Tools for developers to write code. Since then, Copilot has become an additional assistant for Microsoft Bing, Microsoft 365, Dynamics 365, Viva and other services. Today, Microsoft is expanding this ecosystem to include Power BI, Power Pages, Microsoft Fabric, and Windows. It is also launching tools to help developers build their own Copilot, including some new plug-ins that allow Copilot to integrate with the third generation Copilot. Third-party software and service interactions become more useful.

AI plug-ins have only recently become a reality after OpenAI launched the ChatGPT plug-in, and Microsoft introduced the Bing plug-in earlier this month, which allows ChatGPT and Bing Chat to help users perform various tasks, such as finding and booking Dining room.

Microsoft shared in a blog post: "Fundamentally, plug-ins are like a bridge between perhaps a large language model based on public Internet data and what the enterprise may All data kept privately for profit is trained on. This plug-in is a bridge that allows Copilot to access these files when answering questions from enterprise employees."

Now, all Copilot products of Microsoft use The same open plug-in standard that OpenAI created for ChatGPT. Microsoft explains that this way, any plug-in that works with ChatGPT will also be compatible with Microsoft's Copilot.

This also means that developers can use a single platform to develop their own plug-ins for services such as ChatGPT, Bing Chat, Dynamics 365 Copilot, Windows Copilot, etc., so that developers can more easily build a variety of experiences that enable Users will be able to interact with their apps using natural language, Microsoft said.

Improving AI productivity and security

There is no doubt that AI is one of the important themes at this year’s Build conference. After the expanded Copilot and plug-in ecosystem, it is a A series of Azure-related AI enhancements designed to help cloud customers improve productivity and content security.

For example, an update to the Azure OpenAI service, now in preview, will make it easier for customers to add their own data to models built using GPT-4, the most advanced model built by Open AI. Large language models), including a new provisioned throughput SKU with dedicated or reserved capacity and plug-ins for easier integration of external data sources.

Meanwhile, Azure AI Content Safety is a new service that can help enterprises create safer online environments and communities, providing services designed to detect hate speech and violence across multiple languages, text and images. , sexual or self-harm content. Microsoft explains that any objectionable content will be flagged with a severity score, allowing human reviewers to see what content needs urgent attention.

Azure AI Content Safety will integrate various products, including Azure OpenAI Service and Azure Machine Learning prompt flow. Microsoft said the product will be available on June 1.

Other updates include Azure Cognitive Search, currently in preview, which brings vector search capabilities, allowing applications to store, index, and search by concepts and keywords. In addition, developers can build applications to generate personalized responses in natural language, provide product recommendations, detect fraud and identify data patterns, Microsoft said.

Azure Cognitive Search for Language, also now in preview, will help developers interactively customize language skills in their applications, accelerating time to value for organizations looking to use large language models.

Document Translation in Language Studio is now live, a new feature that can batch translate documents, while the Power Automate Translator connector, currently in preview, provides simple workflow automation to translate documents from multiple sources. Text and documents in the cloud, local environment, or local storage.

Putting generative AI into operation

In terms of AI, Microsoft announced a series of updates to its machine learning operating platform Azure Machine Learning, which are said to be designed to make it easier for teams to put generative AI into operation. Responsible generative AI models are put into operation.

Azure Machine Learning include Updates to Prompt Flow include an upcoming preview version of Prompt Flow, providing a streamlined experience for prompting, evaluating, and tuning large language models, which will enable teams to quickly create rapid workflows and integrate these Workflows connect to multiple language models and data sources, using tools to measure their quality.

Now that Microsoft support for base models is also available in preview, Microsoft provides native capabilities to fine-tune and deploy base models from open source repositories using Azure Machine Learning components and pipelines. Other features in preview include support for text and image data in Responsible AI dashboards, allowing users to evaluate large models built with unstructured data during the training and evaluation phases, and a "model monitoring" feature that provides tracking Tools for model performance in production for continuous improvement.

Finally, Microsoft announced various other updates to support large-scale collaboration, governance and rapid development of Azure Machine Learning, including managed function libraries, Microsoft Purview connector, managed network isolation, support for DataRobot 9.0, Azure Machine Learning Registry and Azure Container for PyTorch.

The above is the detailed content of Microsoft expands Copilot and plug-in ecosystem for developers. For more information, please follow other related articles on the PHP Chinese website!