It's been a long time since Google's offline events had such an impact in China, "said a participant at the 2024 Google Developer Conference held in Beijing on August 7-8." Every booth was packed, mainly related to AI
In the current trend of generative AI, Google's domestic conference has also received more attention than ever before. In the exhibition area, from the display of device side generative AI, multimodal search, and the innovation of generative AI in e-commerce to the display of various application cases, including Gemma application intelligent rhubarb, Polyverse 'Gemini 1.5 Pro application display, and golf coaching, each booth is surrounded by people.
Compared to last year's developer conference, this year's exhibition area is mostly generative content. When it comes to this, Chen Junting, President of Google Greater China, said, "This is indeed a key focus of Google, a core task for everyone
The problem of AI is very big. Sometimes AI is not a single new tool, but rather an evolution of existing tools. Therefore, I believe that those who make games, applications, and successful enterprises, as you know, have all evolved their abilities and added AI capabilities Chen Junting stated.
All in AI
In the keynote speech segment, from developer tools to Android and cloud teams, all the products mentioned by the speakers are based on generative AI.
In his opening speech, Chen Junting mentioned the marketing of NetEase Games empowered by Google AI. He mentioned that NetEase Games has improved the efficiency of advertising copy creation overseas by leveraging Google Cloud and Vertex AI platforms. Working hours have been reduced from one week to a few hours, and more than 700 customized copies can be generated per month. In his speech, Chen Junting also mentioned that Xiaomi's new generation flagship device will utilize Google Cloud and Gemini to bring a smarter experience to overseas users.
During his speech, Chen Junting emphasized the strong presence of mobile phone manufacturers at this year's developer conference. OPPO appeared at the conference, and Tang Kai, President of the Software Engineering Division, came to share the innovation brought by generative AI to the mobile phone industry in the global market, as well as OPPO's latest strategy in the AI mobile phone field.
As a revolutionary technology, generative AI has a profound impact on various industries, including the mobile terminal industry. Tang Kai mentioned that OPPO has combined the Google Gemini model to launch a series of innovative AI features in the global market, including AI recording and AI toolboxes.
Among them, AI recording supports multi speaker recognition, which can automatically identify and extract key information, intelligently organize lengthy meeting content, and help users quickly capture and understand the key content in the recording. The AI toolbox integrates three major functions: AI article, AI copy generation, and AI voice reading, helping users efficiently read and produce creative copy. It is reported that these tools are integrated into the ColorOS smart sidebar, which users can easily call at any time.
Tang Kai stated that these AI functions have already been implemented on the all-new ColorOS 14 and have landed on the OPPO Reno12 series models released overseas. He mentioned that through cooperation with partners such as Google, OPPO is vigorously promoting the popularization of AI phones.
Bill Jia, Vice President of Engineering at Google Core ML and Google Cloud, stated at the conference that Google AI is helping OEMs build smarter devices. Google's testing platform Android Device Streaming, in collaboration with mobile phone manufacturers such as Xiaomi, OPPO, OnePlus, Samsung, etc., facilitates developers in terminal testing and is currently in the Beta stage.
Some start-up companies also showcased some practical cases at the conference. Guru Network, a startup focused on going global and gaming applications, is also using Google's Gemma2 model for the podcast application Castbox, allowing users to quickly grasp the key points of content without listening to the entire podcast episode. The AI Mirror application developed by Polyverse optimizes advertising material analysis, copywriting, and creative generation through Gemini's multimodal capabilities, significantly reducing the time and cost of content creation.
Google is very busy this year
The team here has had a quiet a busy year, "said Jeanine Banks, Vice President and General Manager of Developer Relations at Google Developer X, who visited the conference in China for the second year. She shared three different sizes of Google's large models and technological updates at the conference to meet the needs of different developers.
Among Google's large model series, Gemini Nano is the most efficient model that can run directly on mobile devices, providing low latency response and data privacy protection. Gemini 1.5 Flash is Google's fastest and most economical model to date, suitable for handling high-capacity tasks and providing a 1 million token context window. Gemini 1.5 Pro supports a 2 million token context window, making it suitable for complex tasks that require the highest quality response.
In addition, in order to meet the needs of developers for flexibility and control, and to fine tune, enhance, and ground AI according to specific application scenarios, Google has launched the Gemma series open model, which is in line with the Gemini series model. Recently, it officially released the new generation Gemma 2 model, which is more powerful and efficient than the first generation model, and its security has also been significantly improved.
It is reported that Gemini has been integrated into numerous development tools, including Android Studio, Chrome VNet, Project IDX, Colab, VS Code, IntelliJ, and Firebase, which can help developers write, debug, and test code. It can also generate documentation, understand the entire code repository, and become a development assistant.
At the conference, Jeanine Banks mentioned that the supporting application for Xiaomi Auto SU7 is based on the development framework Flutter. On the basis of the original Flutter, Google has launched Flutter 3.24 and Dart 3.5. The biggest highlight of the new version is the early preview version of the new API for Flutter GPU. By using the built-in Flutter SDK, developers can access GPUs using Dart code, thereby improving image rendering capabilities.
Shi Jingyu, the head of the Android Development Technology Promotion Department, shared the latest developments of Mobile at the conference, some of which can enable developers to use Gemini to build the next generation of Android applications. For example, Gemini Nano can run on mobile devices, providing low latency response and data privacy protection, such as implementing intelligent replies in chat applications, while ensuring that data does not leave the device. Even without a cellular network, Gemini Nano can ensure the availability of AI functionality.
This has been an amazing year, "Jeanine Banks said in an interview with First Financial and other media." Last year, when I was in Shanghai, I came into contact with many Chinese startups. At that time, my feeling was that these companies were trying various things to explore AI aspects. A year has passed, and we not only see more AI model constructions and new AI application development, but also see that everyone is developing more complex and sophisticated AI systems
In these complex AI systems, there is not just one intelligent agent, but multiple intelligent agents who work together to complete tasks. Jeanine Banks said, 'It's very gratifying to see so many changes in just one year, and the speed is very impressive.' After one year, she saw a lot of opportunities and a stronger desire to serve the market.
At the same time, Jeanine Banks observed a second significant shift in focus, "Previously, a lot of people's focus was on how to quickly deliver more solutions and applications to more users, which was certainly important. However, now everyone also attaches great importance to one thing, which is that as we scale up AI, we must achieve compliance and fulfill the principles of responsible AI
Chen Junting believes that the progress driven by AI is not limited to cutting-edge productivity, but also has infinite possibilities in promoting cultural inheritance and benefiting society. Guangxi University for Nationalities utilizes Google Global AI open source technology to protect and inherit intangible cultural heritage of ethnic minorities; In the plateau pasture of Sichuan, Southwest University for Nationalities has improved the lives of local herdsmen with the help of Google AI open source technology; Google Philanthropy has partnered with Xinxin Education Foundation to provide AI introductory education to over 900 students in 26 remote primary schools through the "Compile Dreams" project.
When it comes to Google's future plans, Jeanine Banks believes that on the one hand, it is necessary to adhere to open source to help Chinese developers better go global, and on the other hand, it will also use Google's technical expertise to help Chinese developers and development companies face complex overseas business, achieve compliance services for different overseas markets, and adapt products.