Generative AI will profoundly change the way every company operates, marking a new turning point in the development of artificial intelligence technology. Amazon Cloud Technology announced at the New York Summit yesterday that it has launched seven new generative AI functions, further lowering the threshold for using generative AI, so that both business users and developers can benefit from it. With these new features, enterprises from all walks of life can focus more on their core business, improve production efficiency, and fully release the value of data and the potential of generative AI.
Amazon Bedrock fully scaled,
Added new base model, base model supplier and Agents functions
Amazon Cloud Technology announced the comprehensive expansion of its fully managed basic model service Amazon Bedrock, including adding Cohere as a basic model supplier, adding the latest basic models of Anthropic and Stability AI, and releasing the transformative new feature Amazon Bedrock Agents.
Cohere is committed to developing leading enterprise AI platforms and leading-edge underlying models that more intuitively generate, retrieve and summarize information. Anthropic is a research company focused on AI security, building trustworthy, explainable and controllable AI systems. Anthropic has ported its latest language model, Claude 2, to Amazon Bedrock. Stability AI is a community-driven open AI company that provides foundational models that create text, images, audio, video, code, and more from simple text commands. Stability AI will release the latest version of its Vincent graphical model suite, Stable Diffusion XL 1.0 (SDXL 1.0), on Amazon Bedrock.
The Amazon Bedrock Agents feature will help developers easily create fully managed AI Agents. The Amazon Bedrock Agents feature helps businesses accelerate the delivery of generative AI applications that can manage and execute tasks by making API calls to corporate systems. The Amazon Bedrock Agents feature extends the underlying model to understand user requests, break down complex tasks into steps, conduct conversations to gather additional information, and take actions to satisfy user requests. With the Amazon Bedrock Agents feature, users can automate tasks for internal or external customers, such as managing retail orders or processing insurance claims. For example, with the Agents function, a generative AI application serving e-commerce can not only answer simple questions (such as "do you have a blue jacket?"), but also help users complete complex tasks (such as updating orders or managing transactions).
Amazon EC2 P5 instances are officially available to accelerate generative AI and high-performance computing applications
Amazon Cloud Technology announced that Amazon Elastic Compute Cloud (Amazon EC2) P5 instances are officially available. This is a next-generation GPU instance that can meet customers' demands for high performance and high scalability when running artificial intelligence, machine learning, and high-performance computing workloads. The instance is powered by the latest Nvidia H100 Tensor Core GPUs, reducing training time by up to 6x (from days to hours) compared to previous generation GPU-based instances. This performance boost will help customers reduce training costs by up to 40%.
Amazon EC2 P5 instances provide 8 Nvidia H100 Tensor Core GPUs with 640GB of high-bandwidth GPU memory, while providing third-generation AMD EPYC processors, 2TB of system memory, and 30TB of local NVMe storage. Amazon EC2 P5 instances also provide 3200Gbps aggregate network bandwidth and support GPUDirect RDMA, which enables inter-node communication to bypass the CPU, achieving lower latency and efficient scale-out performance.
Vector Engine for Amazon OpenSearch Serverless
Help customers easily build modern generative AI applications
Amazon Cloud Technology announced the launch of the Vector Engine for Amazon OpenSearch Serverless. After being officially available, the vector engine supports simple API calls and can be used to store and query billions of Embeddings. In the future, all Amazon cloud technology databases will have vector functions to help customers simplify operations and facilitate data integration.
Embeddings should be stored close to the source data. A series of factors will affect how enterprises choose the most suitable option. These factors include the current data storage location, familiarity with database technology, the expansion of vector dimensions, the number and performance of Embeddings demand etc. Therefore, Amazon Cloud Technology also provides the following options for more advanced vector data storage needs:
● Amazon Aurora PostgreSQL compatible relational database, supports pgvector open source vector similarity search plug-in;
● Distributed search and analysis service Amazon OpenSearch with k-NN (k nearest neighbor) plugin and vector engine for Amazon OpenSearch Serverless;
● Compatible with PostgreSQL's Amazon RDS (Amazon Relational Database Service) relational database, and supports the pgvector plug-in.
Amazon CodeWhisperer integrates with Amazon Glue to further improve development efficiency
Recently, Amazon Cloud Technology announced that Amazon CodeWhisperer is officially available. This is an AI programming assistant that can use the underlying basic model to help developers improve their work efficiency. It can generate code suggestions in real time based on comments left by developers using natural language and historical code in IDE (Integrated Development Environment). In addition, Amazon Cloud Technology has also released the Amazon CodeWhisperer Jupyter extension to generate real-time, single-line or complete function code suggestions for Jupyter users' Python Notebooks in Jupyter Lab and Amazon SageMaker Studio.
Now, Amazon Cloud Technology officially announces that Amazon Glue Studio Notebooks supports Amazon CodeWhisperer, helping Amazon Glue users optimize their experience and improve development efficiency. With Amazon Glue Studio Notebooks, developers can write specific tasks in natural language (English), such as "create a Spark DataFrame from the contents of a json file". Based on this information, Amazon CodeWhisperer will recommend one or more code snippets that can accomplish this task directly in Notebooks. Developers can choose to "Accept Most Recommended Suggestion", "See More Suggestions" or "Continue to write code yourself".
Amazon QuickSight adds generative BI function to upgrade natural language human-computer interaction
Amazon Cloud Technology announced that it is combining the large language model capabilities provided by Amazon Bedrock with Amazon QuickSight Q, which supports natural language question answering, in order to provide generative BI capabilities in Amazon QuickSight. The feature, which will soon be available on Amazon QuickSight, will help businesses easily explore data, discover and share insights.
With the new generative BI capabilities in Amazon QuickSight, business analysts can use natural language to easily perform everyday tasks, including:
● Create data visualization charts in seconds based on Amazon QuickSight Q’s new visual authoring experience;
● Use natural language to fine-tune and format chart effects;
● Create computing tasks through natural language without learning a specific syntax.
For business users who use dashboards and need to interact with them, Amazon Cloud Technology also released the Stories function to help business users use the powerful capabilities of generative BI to generate, customize, and share highly informative visualizations through natural language prompts chart.
Amazon Entity Resolution is officially available, empowering enterprises to improve data quality and gain customer insights
Amazon Cloud Technology announced the general availability of Amazon Entity Resolution. It is an analytics service powered by machine learning that helps businesses easily analyze, match, and correlate related records that may be stored in applications, different data acquisition channels, and data stores.
Amazon Entity Resolution aggregates customer, business, and product information using rules-based and machine learning-based technologies to customize workflows. Business analysts and developers can quickly improve data accuracy with built-in, preconfigured workflows, or customize workflows to meet enterprise needs. With Amazon Entity Resolution, companies can better understand the association, matching, and connection of data, while digging deeper into customer insights and clearly capturing supply chain data to improve operational capabilities, conduct more effective marketing, and make complex financial investment decisions. Amazon Cloud Technologies also announced plans to add two Amazon Entity Resolution partners, LiveRamp and TransUnion, while enhancing interoperability with the Unified ID 2.0 open source framework. Through these integrations, customers will be able to more easily translate or enrich their records while better protecting information and reducing data movement.
Amazon HealthScribe Uses Generative AI to Help Build Medical Applications
Amazon Web Services announced Amazon HealthScribe, a new HIPAA (Health Insurance Accountability and Protection Act)-compliant service that helps medical software providers build clinical applications. These apps use text recognition and generative AI techniques to generate clinical documentation, saving doctors time.
With Amazon HealthScribe, medical software providers can automatically create reliable records by calling an API, extract key information (such as medical terms and drugs), create summaries from doctor-patient conversations, and then enter this data into electronic health records (EHR) system. Amazon HealthScribe, powered by Amazon Bedrock, enables medical software vendors to more quickly and easily integrate generative AI capabilities into their applications. In general medicine and orthopedics, two common specialties, medical software vendors are already using Amazon HealthScribe, eliminating the need to manage the underlying machine learning infrastructure or train their own medical-specific large language models (LLMs).