What are the must-haves for this Meetup, which brings together the top big names in the industry? | IDP Meetup 01 Highlights Review

On March 5, 2022, the first IDP Meetup was successfully held. With the theme of "AI development and production platform and its co-prosperity ecology", this IDP Meetup brought together 6 top industry leaders to create a wonderful knowledge feast with geeks who are concerned about and love AI and basic technologies.

Quality is King: Gathering Top Celebrities

IDP is committed to creating an unbounded community for free sharing and common communication for geeks who truly care about and love basic technologies. Whether it is for content output or Meetup activities, "high quality" is the core philosophy that the IDP community adheres to.

In this Meetup, we gathered a lineup of guests comparable to the industry's top conferences for topic sharing and roundtable discussions. Participating guests include Liu Jiang, Deputy Dean of Zhiyuan Research Institute, Co-founder of Turing, and former CSDN Editor-in-Chief, Apache Software Foundation Member, former Analysys CTO Guo Wei, former Shell Financial Services Small and Micro Enterprise Ecological CTO Shi Haifeng, Zhangqu Liu Huicheng, Chairman of Technology, Guo Jianjun, founder and CEO of Weiling Times, and Lu Yilei, founder and CEO of Baihai Technology.

The first Meetup event has been widely concerned and loved by AI geeks. The registration participants include algorithm engineers and big data engineers from industries such as finance, mining, and the Internet.

Topic Sharing: Co-Prosperity - AI Development and Production Platform and Its Ecosystem

The topics shared cover cutting-edge topics such as cutting-edge trends in AI research, next-generation AI development and production platforms, AI and the Metaverse, and AI and DataOps.

1. Topic Sharing 1: AI Technology Trends

First, Mr. Liu Jiang, vice president of Zhiyuan Research Institute, made a judgment on the future development trend of AI from the perspective of scientific research and industry. Mr. Liu pointed out that there are three main trends in the future development of AI:

• general large model has become a research hotspot at the forefront of AI. 2020, OpenAI launched GPT-3, setting off a wave of large models. Leading institutions such as Zhiyuan, Huawei, Alibaba, Baidu, Microsoft, and Meta have promoted the research and development of large models. Among them, Zhiyuan Research Institute and "Wu Dao 2.0" released in 2021 are currently the world's largest and most capable ultra-large-scale intelligent model, with a parameter scale of 1.75 trillion, 10 times that of GPT-3, and the world's largest before 1.1x the Google Switch Transformer model.
• AI for Science gave birth to a new paradigm of scientific research . AI is increasingly being widely used in life sciences, chemistry, physics, materials science and other research, innovating traditional research methods and realizing incremental mining.
• AI system has become a battleground for strategists. Since 2012, the computing power of the chip has increased by 7 times, while the demand for computing power by AI has increased by 300,000 times. The huge demand of AI for computing power has made "Sunway TaihuLight", Fuyue and other supercomputers/computing platforms more suitable for AI to flourish. At the same time, under the trend of large-scale application of AI, hardware-based AI systems also show a huge development space.

The development of AI has gone through the era of symbolic AI and perceptual intelligence, and is currently moving towards the era of cognitive intelligence driven by data and knowledge. In addition to the above three major trends, self-supervision, common sense, causality, reinforcement, and brain-like on the technology side, and robots and metaverse on the application side are also important directions that deserve forward-looking attention.

Figure 1. Sharing by Mr. Liu Jiang

2. Topic sharing 2: AI development and production platform trends

Afterwards, Baihai Technology founder and CEO Lu Yilei shared his insights on AI development and production platform products and technologies, covering the trends and classifications of AI development and production platforms and what is the new generation of AI development and production platforms.

In terms of trends, Mr. Lu believes that the development of AI development and production platforms mainly presents three major : one-stop, cloud native and . Among them, simplicity is particularly important. After the evolution from simple to complex integrated tools, simplicity, simplicity and high ease of use will be the future direction of AI development and production tools.

From the perspective of classification, the current AI development and production platforms can be mainly divided into two categories: : One is the AI basic software platform represented by IDP, JupyterLab, etc. Its core features are user-oriented, on-demand use, professional light The other type is an integrated algorithm development platform represented by products of cloud manufacturers. Its core features are guided by the whole process of AI development and production, and comprehensively integrate tools in all links, which is heavy and complex.

Finally, Mr. Lu introduced the new generation of AI development and production platform - IDP (Intelligent Development Platform) launched by Baihai Technology from the aspects of value proposition, product and technical architecture and typical application scenarios. IDP complies with the three core trends of AI development and production platforms mentioned above, adopts a cloud-native architecture, uses IDE as the only entry point, and uses silk-smooth built-in plug-ins to meet the one-stop requirements of algorithm developers for data access, model Requirements for the whole process of development, training, and release. For users such as algorithm engineers and data scientists, the core value of IDP is professionalism, convenience, and ease of use, which can significantly improve their work efficiency; for enterprise customers, IDP provides high-performance computing engines and cross-team collaboration to help. Enterprises reduce costs and increase efficiency, and accelerate innovation.

Figure 2. Sharing by Mr. Lu Yilei

3. Topic Sharing 3: The Collision of AI and Cloud Games

Guo Jianjun, founder and CEO of Weiling Times, discussed with everyone how AI technology will reshape cloud games under the metaverse trend.

Teacher Guo Jianjun first shared his understanding of the Metaverse. In the era of the Internet and mobile Internet, the foundation is computing. The future of the original universe is visualized, and its basis will change from computing to computing plus rendering, especially real-time rendering . The metaverse is a process in which carbon-based organisms start from carbon-based continents, cross the entire sea of computing power, and find silicon-based continents. The aborigines of the metaverse are digital people, and the rules of the metaverse are the rules of AI.

Games are the best way for us to understand the metaverse, because all interactions in games are based on real-time rendering. The production of games will generate a lot of demand for AI. The main demand points include:

• AI Content Generation (AIGC): implementing a large scene or a large model in the game, it is very difficult to rely solely on the game designer, so more AI-driven software is needed to assist the generation of content, And through the learning ability of AI, more "custom" scenes can be generated independently.
• AI image quality supplement: extensive use of AI capabilities for image quality enhancement and supplementation, rather than relying solely on rendering and encoding. While improving image quality, it can effectively reduce costs.
• AI playing games : With the help of AI computing power, NPCs in future games will be more intelligent and vivid, and can replace "players" to take over the adventures of characters in the game world.

The realization of these scenarios requires the cooperation of the entire industry and even cross-industry, including AI development and production platforms.

Figure 3. Sharing by Mr. Guo Jianjun

4. Topic Sharing 4: Collaboration between AI and DataOps

Finally, Mr. Guo Wei, ASF Member and former Analysys CTO, shared his insights on AI and DataOps.

Driven by digital and intelligent transformation, various 'Ops' have exploded in recent years. Based on his deep experience in DataOps and open source fields, Mr. Guo observed that there are three major trends in Ops:

• data volume and data complexity have increased, and the user group scenarios have become increasingly complex. predicts that the compound annual growth rate of China's data volume will reach 24% in the next five years. Various data sources, data technologies, and data quality have increased the complexity of data. On the user side, the demand scenarios for data are increasingly diversified, such as fragmentation, real-time, efficiency, real-time interaction, etc.
• Chinese scenes are rich, while overseas scenes are relatively focused. China's data application scenarios are extremely rich, and the products usually show the characteristics of "big and comprehensive". While overseas products or open source projects are mostly focused on a specific scenario and small problem, they are deeply cultivated and specialized.
• China's open source is technology-driven, while overseas open source is commercial open source . Compared with overseas, China's open source business is still in its infancy, and it is still mainly driven by technology. Most of them are projects built by geeks to "glow for love". Overseas awareness and acceptance of open source has become relatively mature, and open source projects have entered the commercialization stage.

Mr. Guo took DolphinScheduler as an example to explain the technical map and application value of DataOps. The core goal of DataOps is to improve data usage efficiency and lower the threshold for data usage. In the future, AI will be the main technology for data mining and utilization. The synergy of AI and DataOps will further promote enterprises to mine and use data faster, better, and more deeply, and accelerate digital and intelligent transformation and innovation of enterprises.

Figure 4. Sharing by Mr. Guo Wei

Roundtable Discussion: Creating the Future - Prospects for AI and AI Development and Production Platforms

After the theme sharing, the guest teachers had a round-table discussion on the future trends of AI and AI development and production platforms, and put forward thought-provoking and wonderful views.

Figure 5. Round table sharing

(From left to right: Teacher Liu Jiang, Teacher Guo Huicheng, Teacher Guo Wei, Teacher Shi Haifeng, Teacher Lu Yilei)

Guo Huicheng, chairman of Palm Fun Technology, pointed out from the perspective of the demand side that the core factor restricting the large-scale use of AI by enterprises is that it is difficult to measure the value of AI and AI development and production tools to the business. As a production operator, it is difficult to make large-scale investment in high computing power and AI applications without knowing what benefits AI can bring to business operations.

Mr. Guo Wei proposed the "four-quadrant diagram" theory of the platform tool product market: the top of the vertical axis is large enterprises, the bottom is small enterprises, the left side of the horizontal axis is to help enterprises improve efficiency, and the right side is to help enterprises make money. The ideal market position is to help large enterprises improve efficiency and help small enterprises make money. Big companies usually do things to make money by themselves, while small companies mainly solve survival problems, and improving efficiency is not their primary pain point.

Teacher Shi Haifeng said that the ultimate trend of any technology, including AI development and production platforms, is to make it easier for more people to enter the industry. "The product is difficult to get started with a moat" is actually a misunderstanding. The future trend is that we can find tools that are easier to use and easy to use, and can obtain various guarantees after use, including consulting services, purchase of commercial versions, etc. This is a healthy and complete ecosystem.

Teacher Liu Jiang said that there will be broad demand for AI development and production tools. As the industry flourishes and matures, it is an inevitable business law to generate large-scale demand for tools and platforms. At present, it is a good time for the development of Chinese technology enterprises, and it is also a golden age for domestic substitution.

Based on the personal experience of IDP's practical exploration, Mr. Lu Yilei shared his views on the evolution of AI development and production platform product route and business model. From the perspective of product route, the AI development and production platform must first solve the problem of AI application from 0-1, so that everyone can use it conveniently. In the future, with the increase of algorithm engineers, the large-scale application problems of 1-100 or even 10,000 will be gradually solved. Judging from his 15-year experience in big data, Mr. Lu predicts that 70-80% of big data engineers will be transformed into algorithm engineers in the future. From the perspective of business models, AI development and production platforms must be able to help enterprises solve practical business problems. At present, the main needs of enterprises are to quickly open resources and quickly conduct model training, and second, to go deep into business scenarios to solve specific problems.

More exciting next issue

Starting with this Meetup, we will regularly invite big names in the industry to conduct in-depth exchanges and discussions with you on topics such as AI and basic software technologies and applications. Welcome to follow Baihai IDP to get the latest information on community activities. IDP looks forward to seeing you next time!

[Easter eggs! ! ]

Follow the Baihai IDP official account and reply to [IDP Meetup] to receive the sharing materials of "Theme Sharing-AI Technology Trends" and "Theme Sharing-New Generation AI Development and Production Platform" for this IDP Meetup.