Google deeply integrates generative AI capabilities into its cloud platform
Help overseas companies and developers
Easily leverage AI to drive and achieve innovation
Let’s take a look at the 2023 Google Developer Conference
How Google Cloud accelerates innovation and enhances information security
Pay attention to Google Cloud’s 3 key AI development areas
Learn how generative AI capabilities can help bring ideas to life
Overview of the wonderful conference site
AI remains at the core of everything Google Cloud does. AI products and models are Google Cloud's first key AI development areas, helping overseas companies and developers to easily build and deploy generative AI industry applications to ensure enterprise-level security and privacy. Google Cloud makes it easy for enterprises to leverage AI to drive and achieve radical innovation.
Vertex AI is Google Cloud's end-to-end machine learning platform. It is the most comprehensive enterprise-level generative AI platform on the market today, helping developers accelerate the development and deployment of machine learning models. It brings together Google's most cutting-edge "black technology": rigorous AI research results, proven AI technology and basic models, and a comprehensive AI tool suite.
Vertex AI
At present, Google has launched multiple series of basic models, covering a variety of different interaction modes, from text chat to images, code, voice and Embeddings. Now, these base models are getting a complete upgrade.
Basic model provided by Google
In the new version of the text model , the model's input length limit increases from 8,000 tokens to 32,000 tokens. This allows developers to easily process longer documents such as academic papers, legal documents, books, and more. PaLM 2 now officially supports 38 languages, including Chinese, Japanese, German, Spanish, and more. An additional 100 languages are in closed preview.
PaLM is used for chatting and text editing
The text-to-code model Codey has also received an upgrade. In the new version, the quality of code generation and code chat has been improved by nearly 25% for the supported mainstream languages.
Image model Imagen has added a "style adjustment" function to its existing image generation, editing, annotation and visual question and answer functions, allowing developers to only provide about 10 reference images to generate brand features or other creative ideas. Required pictures.
Adhering to Google's consistent philosophy of building an open ecosystem, we have further enriched Vertex AI's Model Garden model library , adding Meta's LLaMA 2 model and TII's Falcon model, and will soon add Anthropic's Claude 2 model. In this way, Google Cloud includes Google's own models, third-party models and various open source models on one platform.
Vertex AI currently has a selection of more than 100 models, allowing developers to freely select the most appropriate model based on factors such as interaction mode, model size, performance, and cost.
Open model ecology
In order to improve the practicality of the model, we also launched Vertex AI Extensions to open the pipeline between the model and real-time data, and allow the model to perform various real-world operations.
Vertex AI extension
To meet the two most common application scenarios of generative AI: search and chat, Google launched Vertex AI Search and Conversation, which complement the basic model and help developers build more powerful search and chat applications more efficiently.
Vertex AI Search can be viewed as a Google search engine that only covers enterprise-owned data, allowing enterprises to provide users with the same level of generative AI experience as Google search. This product applies the most advanced information retrieval and in-depth retrieval technology to enterprise data, not only providing a convenient experience out of the box, but also greatly lowering the application threshold.
Vertex AI Conversation significantly reduces the difficulty of developing safe, practical, and lifelike chatbots, allowing developers to easily build and deploy text and speech chatbots.
Information security is Google Cloud's second AI focus development area. Google provides comprehensive, minute-level front-line threat intelligence and security vulnerability information, and uses technology to help enterprises maintain model security within their own security boundaries to help protect their data and IP.
Google has developed Security AI Workbench based on the existing information security platform. It is the first AI-based platform in the industry and has good scalability. It is based on a large-scale language model dedicated to security purposes, called Sec-PaLM , which can provide assistance to information security teams, analyze the attack paths of threat sources, generate case investigation reports, and continuously monitor various information security threats to Help enterprises avoid becoming the next target of attacks.
In order to seize the opportunities brought by AI, the third development focus of Google Cloud is Google Cloud's platform interface and infrastructure optimized for AI .
Duet AI uses generative AI technology and can be used as a professional AI assistant through different Google Cloud interfaces or interfaces to provide assistance in application development, operation and maintenance, data and security threat identification, reducing the burden and increasing efficiency of the team. Developers can also use Duet AI gets the code explanations you need directly in the interface, simplifying the maintenance of large code bases.
Duet AI on Google Cloud
Start leveraging AI easily with Google Cloud now
Drive and enable innovation
Stay tuned for the 2023 Google Developers Conference
More exciting content updates
Special speeches are being updated continuously
Scan the QR code to go directly to the official website
Explore more highlights from the conference
Follow the "Google Developer" WeChat official account to get more conference information
Bun releases official version 1.0, a magical bug in Windows File Explorer when JavaScript is runtime written by Zig , improves performance in one second JetBrains releases Rust IDE: RustRover PHP latest statistics: market share exceeds 70%, the king of CMS transplants Python programs To Mojo, the performance is increased by 250 times and the speed is faster than C. The performance of .NET 8 is greatly improved, far ahead of .NET 7. Comparison of the three major runtimes of JS: Deno, Bun and Node.js Visual Studio Code 1.82 NetEase Fuxi responded to employees "due to BUG was threatened by HR and passed away. The Unity engine will charge based on the number of game installations (runtime fee) starting next year.