On July 7, the 2023 World Artificial Intelligence Conference Tencent Forum was held at the Shanghai World Expo Center. The theme of the forum is "Infinite Possibilities", and Tencent announced a series of AI innovation progress.
In the field of AI large-scale models that are widely concerned by the industry, Tencent Cloud announced the upgrade of the three major areas of technology base, innovation scenarios, and AI for Science. Among them, the latest upgrades of the self-developed Xingmai high-performance computing network and vector database will provide more powerful underlying capabilities for large-scale model industry applications; innovative scenarios such as financial risk control, interactive translation, and digital intelligence will provide a new foundation for the industry's large-scale model industry. Exploration provides more possibilities; AI for Science continues to apply the latest technologies such as AI large models in the fields of astronomical exploration and cultural archaeology, and uses technology to promote sustainable social value innovation.
Li Qiang, vice president of Tencent Group and president of government and enterprise business, said: In the era of large-scale models, data, network, and computing power constitute the "iron triangle" of the underlying infrastructure. In addition to providing MaaS services, Tencent Cloud is also providing cloud services for the industry. The creation of large models provides infrastructure services such as HCC high-performance computing clusters and vector databases.
Wu Yunsheng, vice president of Tencent Cloud, head of Tencent Cloud Intelligence, and head of Youtu Lab, believes that the development of AI large-scale model technology and industrial exploration are inseparable from industrial chain coordination and ecological co-construction. Tencent hopes to join hands with industry partners to jointly promote large-scale The innovation and implementation of the model in the industrial field.
[Vice President of Tencent Cloud, Director of Tencent Cloud Intelligence, and Director of Youtu Lab Wu Yunsheng]
At this conference, Tencent, as a representative large-scale model enterprise, joined the large-scale model ecological cooperation community initiated and established by the China Academy of Information and Communications Technology. In the future, it will support large-scale model open service platforms and industry-university-research exchanges.
Tencent Cloud MaaS has been upgraded again to implement innovative scenarios such as financial risk control, interactive translation, and smart people
The ability to quickly understand and apply large models is the key to maintaining the core competitiveness of the market in the new wave of artificial intelligence technology, and this has become the direction that more and more companies are exploring.
In order to solve the practical challenges of enterprise application model technology, such as limited computing resources, poor data quality, high input costs, and little professional experience, Tencent Cloud first announced the research and development progress of the industry's large model on June 19, relying on the Tencent Cloud TI platform to build The select store of large-scale models in the industry provides customers with MaaS (Model-as-a-Service) one-stop service to help customers build exclusive large-scale models and intelligent applications. At present, Tencent Cloud has provided more than 50 large-scale industry solutions for more than 10 industries including media, cultural tourism, government affairs, and finance.
At this World Artificial Intelligence Conference, Tencent Cloud once again announced a number of upgrades.
Among them, in the field of technology base, the self-developed Xingmai high-performance computing network and vector database provide more abundant computing power infrastructure for large-scale industrial applications. Among them, the newly upgraded Tencent Cloud self-developed Xingmai high-performance computing network can increase GPU utilization by 40%, save 30% to 60% of model training costs, and bring a 10-fold improvement in communication performance for large AI models. Based on Tencent Cloud's new-generation computing power cluster HCC, it can support a super-large computing scale of 100,000 cards. Tencent Cloud's AI native vector database supports up to 1 billion-level vector retrieval scale, and the delay is controlled at the millisecond level. Compared with the traditional stand-alone plug-in database, the retrieval scale is 10 times higher, and it also has a peak capability of millions of queries per second (QPS).
In terms of application innovation, Tencent Cloud's large-scale industry model capabilities have been applied to scenarios such as financial risk control, interactive translation, and digital smart customer service, which has greatly improved the efficiency of intelligent applications.
The financial risk control solution supported by the industry's large-scale model has 10 times the efficiency compared with the previous one. Through Tencent's accumulation of more than 20 years of black and gray production confrontation experience and thousands of real business scenarios, the overall anti-fraud effect is 20 times higher than the traditional model. An increase of about %. Based on the prompt mode, enterprises can iterate risk control capabilities, from sample collection, model training to deployment and launch, to achieve zero manual participation in the whole process, and the modeling time is reduced from 2 weeks to only 2 days. Even with limited sample accumulation, rapid construction can be completed and the "cold start" process can be skipped.
In the field of interactive translation, based on the industry's large-scale model technology, simultaneous interpretation technology no longer requires millions of training data, and only needs "small sample" training to achieve better results. Translation in professional fields can also reduce the need for manual tuning. Participate, guarantee the translation effect, and land in multiple vertical industries. Among them, Tencent simultaneous interpretation has provided AI simultaneous interpretation services for the main forum of the World Artificial Intelligence Conference for six consecutive years.
In the field of digital humans, Tencent Cloud launched a small-sample digital human factory this year, which can reproduce 2D digital clones within 24 hours with only a small amount of data, greatly reducing the cost of enterprise application digital human services. Now, relying on the AI generation algorithm, the reproduction speed of the 3D image of the digital intelligent person has been greatly improved. Driven by generative actions, combined with the ability of the industry's large model, the enterprise can obtain more "personalized, professional, natural and realistic" digital intelligent employees , making "face-to-face" professional services possible.
A new breakthrough in AI for Science, the star exploration plan uses cloud + AI to detect 2 fast radio storms
The application of Tencent Cloud's large-scale model in the industry, in addition to exerting an effect on the digital transformation of the industry, also accelerates the application of large-scale model and other AI technologies in the field of scientific computing, and explores more possibilities of "technology for good".
In 2021, Tencent, the National Astronomical Observatory, and the School of Computer Science and Technology of Fudan University will jointly launch the "Star Exploration Project". Based on the leading computer vision technology of Tencent Youtu Lab and the leading computing power of Tencent Cloud, all parties will use cloud + AI to help China Sky Eye FAST processes the huge amount of data received every day, and finds fast radio bursts and pulsar clues through visual AI analysis. Up to now, 30 pulsars have been discovered. This year, the star exploration program has made progress again, discovering two fast radio bursts for the first time through AI technology.
Compared with pulsar exploration, in order to discover fast radio bursts that occur at a lower frequency in massive data, AI models are required to have higher accuracy and faster calculation speed. In order to improve the calculation speed, Tencent specially designed a set of brand-new, end-to-end AI algorithms for the exploration of fast radio bursts. Under the same computing power, this brand-new astronomical data processing paradigm promotes the signal processing efficiency to be 1800 times faster than the conventional processing process. At the same time, combined with AI training assisted by Tencent Cloud's industry large model, through self-supervised pre-training + data fine-tuning, the accuracy of the model is improved and the cost of data labeling is reduced.
In the field of culture, Tencent has innovatively applied AI algorithms such as text detection, copy generation, and font matching to the research of oracle bone inscriptions. Based on the data annotation and processing of "Yinqi Wenyuan" oracle bone inscriptions, the oracle bone inscription model library has been continuously enriched and improved through customized algorithms. Up to now, the world's largest oracle bone inscription single-character database covering 1.43 million characters has been established to improve the recognition and interpretation of oracle bone inscriptions, as well as the content of oracle bone inscriptions Efficiency of extraction etc. By exploring the new model of man-machine collaboration researched by Oracle, Cloud + AI will further revitalize the source of Chinese characters and show the infinite charm of Oracle to more people.
Released the "Report on Ten Trends of AI in the Era of Large-scale Models", and large-scale model technology drives "human-machine symbiosis"
In response to the development of large-scale AI technology, Tencent Research Institute, Tongji University, and Tencent Cloud jointly released the "Human-Machine Symbiosis—Ten Trends Report on AI in the Era of Large-scale Models" based on the research of three parties in the field of AI. From the perspective of society, etc., the observation of key trends in the era of large models is proposed.
[Tencent Research Institute, Tongji University, and Tencent Cloud jointly released "Human-Machine Symbiosis - Ten Trends Report on AI in the Era of Large Models"]
The report points out that emergence and fusion are important AI trends in the era of large models. Large models push AI into the AGI stage, while the development of multimodal AI provides the ability to solve complex problems and realize multimodal intelligent processing. These trends bring new hopes and opportunities for the rapid evolution of artificial intelligence and the realization of AGI goals.
With the advent of the era of large models, the human-computer interaction mode has undergone revolutionary changes, bringing unprecedented empowerment to individual creators. On the one hand, the emergence of the plug-in mechanism enables the platform to open the "application moment", which brings great possibilities for the scene application of the model. A digital human with a more "soul" is also expected to become the entrance of the next generation of human-computer interaction. The innovation of human-computer interaction will empower individuals to become super producers.
In terms of AI-assisted industrial development, MaaS services and vertical field applications will be important trends in the development of AI in the era of large-scale models. By building a controllable and usable security ecosystem and promoting the implementation and application of models, AI technology will bring benefits to various industries. More opportunities and challenges. The innovation and development brought about by the era of large models will push artificial intelligence into a broader future.
Si Xiao, President of Tencent Research Institute, said in the preface of the report that with the development of AI, it has penetrated into other industries, and an era of national productivity explosion seems to have ushered in the dawn. Whether it is an ordinary individual creator, or an enterprise manager or decision-maker, grasping the perception of AI and judging the trend has become a ticket to the future. The report hopes to provide a comprehensive, objective and in-depth perspective to promote rational thinking and active exploration of the development and application of AI technology.
Technological development is the source of living water for industrial innovation, and the ultimate goal of technological application is to serve people. Tencent continues to invest in advanced technology research and development, transforming sophisticated and complex cutting-edge algorithms and theories into low-threshold "AI applications" and "AI solutions", from Sapiens, TI platforms to MaaS solutions, activating the innovation momentum of the real industry. In the future, Tencent will continue to focus on the new training ground of "industry", and comprehensively promote the innovation and popularization of AI technology from industrial scenarios to research fields, so that "affordable" and "understandable" AI can benefit more scenarios, wider users.