At present, the implementation of generative AI applications has moved from a hot "battle of hundreds of models" to a new stage in which a few excellent models stand out, tool chains are flourishing, and business executives are seriously looking for generative AI implementation scenarios. Many people are asking: "Is this basic model suitable for me? Can I make it more suitable for my field? How can I innovate my IT architecture to quickly build AI Native applications? How to measure the value of introducing generative AI?"
Based on this background, Amazon Cloud Technology specially held the Amazon Cloud Technology Generative AI Builders Conference, which concluded successfully on the afternoon of October 24. In this conference, industry leaders and technical experts focused deeply on the cutting-edge technology of generative AI, conducted in-depth sharing and exchanges on the hot technical topics and popular application scenarios of generative AI, and explained to developers how to use it now. Respond to the opportunities brought by generative AI and maintain strong competitiveness in the AI era.
Accelerate big data and AI to benefit everyone
Amazon Cloud Technology helps enterprise builders unleash the potential of generative AI
At the beginning of the conference, Chen Xiaojian, general manager of the product department of Amazon Cloud Technology Greater China, pointed out in a keynote speech titled " Empowering the new era of generative AI and promoting the inclusiveness of data and AI" that generative AI has now become a key technology for all industries and industries. The primary focus of organizational business leaders. The entire application of generative AI is like an iceberg floating on the sea. The basic model that people often mention is just the tip of the iceberg that we can see. At the bottom of the glacier, a large number of services other than the basic model are also needed to support it, such as Accelerate chips, databases, data management, application integration, etc.
Amazon Cloud Technology provides users with a complete end-to-end generative AI technology stack. From the bottom acceleration layer such as acceleration chips and storage optimization, to the middle layer model building tools and services, to the top layer generative AI-related applications, each layer continues to innovate according to the different needs of customers.
As a leader and innovator in the field of cloud computing, generative AI has always been the main direction of Amazon Cloud Technology's focus and investment. Promoting the universalization of generative AI and helping more enterprises and individual developers accelerate innovation is one of the missions of Amazon Cloud Technology.
In practice, Amazon Cloud Technology has noticed that the business value brought to enterprises by typical application scenarios of generative AI is mainly reflected in three aspects: enhancing customer experience, improving employee productivity and creativity, and optimizing business processes. Based on this, Amazon Cloud Technology helps enterprises and developers innovate business models starting from typical scenarios and quickly build cost-effective generative AI applications to create differentiated competitive advantages and accelerate agile business innovation.
In order to allow customers to easily and securely choose a basic model on demand, Amazon Cloud Technology has previously launched two services , Amazon Bedrock and Amazon SageMaker JumpStart . Amazon Bedrock is the easiest way for enterprises to build and scale generative AI applications using foundational models, providing broad model selection, data privacy, and the ability to customize models without managing any infrastructure. With Amazon Bedrock, users can easily leverage their organization's data as a strategic asset, use custom underlying models, and create more differentiated advantages. In addition, Amazon Bedrock also fully considers user security and privacy protection to help customers protect sensitive data.
Compared with Amazon Bedrock, Amazon SageMaker Jumpstart provides more flexible and customized model fine-tuning. Users can select an open source basic model from SageMaker Jumpstart, then choose different methods according to their own needs, further determine the fine-tuning framework, and use distributed training to implement fine-tuning to better evaluate the fine-tuning effect.
Kingsoft Office: Amazon Bedrock helps WP S
Accelerate building generative AI applications
In the theme sharing, Liu Qiang, AI R&D Director from Kingsoft Office Software Co., Ltd., shared how Kingsoft Office uses generative AI and Amazon Bedrock to help enterprises simplify the process from unified data to personalized customer experience.
Liu Qiang mentioned that starting this year, WPS began to fully introduce the capabilities of large language models into products and is committed to developing a new generation of office software. In the process of building generative AI applications, the limited performance of the basic model, the difficulty in ensuring data privacy and security, and high management costs have become numerous obstacles for Kingsoft Office. As one of the first trial users of Amazon Bedrock, Kingsoft Office immediately started docking and testing with Amazon Bedrock.
The leading large language model supported in Amazon Bedrock meets the needs of Kingsoft Office in multiple word processing scenarios. In addition to a variety of model options, Amazon Bedrock also provides full support for Kingsoft Office at the data security level , greatly improving Kingsoft Office's internal development efficiency and helping Kingsoft Office further innovate the office experience.
Cost-effective infrastructure is the key to building generative AI applications. In the following sharing, Chen Xiaojian mentioned that for more than ten years, Amazon Cloud Technology has been investing in self-developed chips with its partners to provide users with a wide range of high-performance, low-cost machine learning infrastructure options to help customers Accelerate the implementation of generative AI.
In addition, Amazon Cloud Technology recently announced the official launch of Amazon EC2 Inf2 instances. This instance is powered by Inferentia2, the latest Amazon Cloud Technology-designed machine learning chip. Compared with other similar EC2 instances, Inf2 instances are 40% more cost-effective in inference, and Trn1 instances save up to 50% in training costs. Based on this, Amazon Cloud Technology provides high-performance and significant cost-saving instances for deep learning training and inference.
At the data strategy level, from storing, querying and analyzing data to making data useful through business intelligence, machine learning and generative AI, Amazon Cloud Technology provides a broad set of tools to support the entire end-to-end data process. Over the years, Amazon Cloud Technology has achieved certain results in Zero ETL by deepening the integration between services. These deep data integrations between Amazon Cloud Technology services are helping enterprises perform analytics and machine learning without moving. data.
In the current situation of data explosion, data governance is crucial for enterprises. Amazon DataZone's new data governance service supports data discovery, management of cross-departmental access and usage lifecycle across enterprise organizations. Through a unified data directory and a clear division of data sovereignty, data users in the enterprise can easily discover valid data. At the same time, data providers can strictly manage data visibility and access authorization, so that only the right people are allowed to access the right data. The data. This capability is particularly important for today's enterprises that need to build generative AI responsibly, strictly protect enterprise data, and correctly leverage the value of enterprise private data.
Siemens: Reshape the value of data and promote business innovation with the power of generative AI
Li Chaoming, head of IT data analysis and artificial intelligence of Siemens Group, shared with everyone in the theme sharing how Siemens works with Amazon Cloud Technology to leverage the value of enterprise data and build enterprise knowledge based on generative AI through the comprehensive improvement of AI and data capabilities. Chatbot.
He proposed that the current development of generative AI has brought revolutionary changes to all walks of life. In the process of Siemens promoting the implementation of generative AI, Siemens IT DA team has had long-term and in-depth cooperation with Amazon Cloud Technology. Faced with the business problems faced by Siemens China IT such as dispersed information within the enterprise, insufficient data information transmission capabilities, and lack of knowledge operation capabilities, Amazon Cloud Technology worked together to create "Xiaoyu", an integrated solution for Siemens China's business knowledge, which greatly improved Siemens' Internal team productivity and innovation effectiveness.
The emergence of generative AI has brought new opportunities and challenges to the traditional cloud-native application development architecture. With the help of cloud-native services, we focus on a design framework with microservices and event-driven architecture as the core. Serverless gives priority to data insights as the core. Security and data compliance and protection, using modern application governance concepts such as DevOps to build an efficient and agile builder culture, etc., to accelerate the construction and innovation of cloud-native to AI-native applications.
Amazon Cloud Technology foresees that programming will be one of the areas where generative AI technology will be rapidly applied. Amazon CodeWhisperer provided by Amazon Cloud Technology is an artificial intelligence coding partner that can be used in the advanced options of the basic model to generate code suggestions in real time, fundamentally improving developer productivity. The "2023 AI Developer Ecosystem Report" also pointed out that Amazon CodeWhisperer currently leads the field of domestic AI programming tools and is the AI programming tool with the fastest growth rate of domestic users.
Covestro China: Significantly improve development efficiency with Amazon CodeWhisperer
Yu Zhou, head of the China Cloud Development and Operations Team at Covestro, shared with everyone how they improve code development efficiency and accelerate the construction of generative AI applications based on Amazon CodeWhisperer.
Covestro China has two core tasks in daily development: project framework construction and stock code optimization. In practice, repetitive, mechanized paradigm generation, idiomatic algorithm references and enterprise-level API interface customization are the main challenges faced by Covestro China. With the support of Amazon CodeWhisperer, Covestro China's daily code development efficiency and the accuracy of its open source code application tracker have been significantly improved, playing an important role in the process of building cloud native product applications.
In the following sharing, Chen Xiaojian also specifically mentioned that Amazon Cloud Technology has been continuously helping the Chinese developer community to innovate rapidly. Since this year, Amazon Cloud Technology has launched the Amazon Cloud Technology Entrepreneurship Accelerator for startups . In September this year, Amazon Cloud Technology teamed up with the guidance of the Beijing Municipal Education Commission, the Beijing Science and Education Development Foundation, the Basic Model Research Center of the Tsinghua University Artificial Intelligence Institute, as well as customers and partners such as Wall-Facing Intelligence, Zhipu AI, and Baichuan Intelligence to jointly The "First China Generative AI Innovation Application Challenge" was released . Today, the efforts and passion of developers are changing the world and constantly expanding the boundaries of technology. Amazon Cloud Technology will continue to empower builders and continue to work with partners and startup circles to expand the "circle of friends" of generative AI. Further promote the widespread implementation of generative AI.
Summit forum discusses practical experience
Industry leaders share a look at the present and future of generative AI
In the roundtable discussion titled "Let's talk about the present and future of generative AI", Lepton AI CEO Jia Yangqing, Baichuan Intelligence co-founder COO Ru Liyun, Aishi Technology CEO Wang Changhu, Amazon Cloud Technology Greater China Product Department President Manager Chen Xiaojian and Hu Ying, Director of Startup Ecosystem and Investment Business Development of Amazon Cloud Technology Greater China, conducted an in-depth discussion on the opportunities and challenges faced by developers in the new stage of generative AI application implementation.
During the exchange and sharing, industry leaders shared their views on how to choose a basic model, how to build a data strategy in the generative AI era, how to use cloud native to accelerate the construction of AI Native applications, and what development tools are available to accelerate the construction and implementation of generative AI applications. Many developers have made valuable suggestions.
Interpretation of generative AI application secrets in sub-forums
Amazon Cloud Technology helps create best practices for generative AI applications
Developing generative AI applications is a systematic project full of challenges. It is not just a splicing of products and services. Therefore, in order to accelerate the commercialization of customers' final applications, in addition to its own resources, Amazon Cloud Technology also needs to build strong partners. "Friends Circle" works together to help solve various technical problems in the construction of generative AI applications and accelerate the implementation of applications.
In order to explain in more detail the methodology of how Amazon Cloud Technology helps builders unleash the potential of generative AI in view of the critical path to building generative AI applications end-to-end, the sub-forums focused on "Basic Model Priority" - Reconstructing Applications in the Generative AI Era Base", creating a data strategy in the generative AI era, rapid AI native applications based on cloud native services and other hot topics. Technical experts from Amazon Cloud Technology were invited to join hands with customer and partner representatives to give a rich and comprehensive sharing. From theory to practice, it should inspire builders to develop generative AI quickly, safely and efficiently.
In the "Developer Forum" , Amazon Cloud Technology invited well-known industry experts to explain and share with everyone. Let’s take a look at their experiences and insights!
Developer Forum - Helping developers innovate and build in the era of generative AI
In 1024, a developer-only festival, the conference also set up a developer forum to invite industry developers and experts to share the latest development technologies, tools and implementation practices of generative AI. Open source AI algorithm engineer and bilibili's well-known technology UP Zhang Zihao (@同记子浩哥) shared "N Ways for Individual Developers to Play with Generative AI". Zheng Yubin, senior developer evangelist of Amazon Cloud Technology, shared Amazon CodeWhisperer, the key technology of generative AI for developers. Huang Haowen, senior developer evangelist of Amazon Cloud Technology, shared the practice of efficient development of generative AI applications based on large models. Zilliz Partner Luan Xiaofan, technical director and technical advisory committee member of LF AI & Data Foundation, introduced the new infrastructure in the AIGC era - vector database and its application scenarios. Cai Truffle, co-founder & CTO of ApeCloud, shared how to use generative AI on K8S. Gu Si, the chief developer evangelist of NebulaGraph, introduced in detail the graph-based knowledge structure to break through more challenges in the RAG field.
Hands-on training camp—the speed and passion of development in the generative AI era
We have simultaneously prepared hands-on experiments for developers, including Amazon DeepRacer reinforcement learning and autonomous driving model practice, real-time AI programming assistant Amazon CodeWhisperper practice, Amazon Bedrock first experience, and building intelligent
Amazon Cloud Technology provides a complete end-to-end generative AI technology stack, from the underlying acceleration chip and storage optimization, to the middle-layer model building tools and services, to the top-level generative AI-related applications . Promoting the universalization of generative AI and helping more enterprises and individual developers accelerate innovation is one of the missions of Amazon Cloud Technology. Amazon Cloud Technology is willing to work hand in hand with enterprise builders and individual developers to lead the future of artificial intelligence.
Click " Read the original text " now to learn more about the Amazon Cloud Technology Generative AI Builders Conference !