2023 World Artificial Intelligence Conference： Thermal Collective Release of ＂Big Model＂
The highlight of this year’s WAIC is the "big model", and the exhibition site also gathered key "guests" in the field of big models. At this conference, more than 30 basic large models and vertical large models were collectively unveiled, some of which were aimed at general use and some were deeply cultivated in the industry, further promoting artificial intelligence from perception to cognition, from recognition to generation, and from special use to general use.
Every year, the artificial intelligence conference is a stage for technology companies to show their hard power. At this year’s conference site, companies focus on the evolution of AIGC technology, application scenarios, future trends, and standard governance of the whole industry chain, and show the large model to empower the diverse ecology of all walks of life around AIGC+ financial, medical, media, education and other diverse demonstration scenarios and popular industry applications, so that the public can see the practical charm of technology.
IFLYTEK Spark model becomes the hottest "punch point"
During the conference, a variety of the latest industry application achievements equipped with the Spark Cognitive Model entered the audience’s field of vision at the Iflytek booth. The digital press officer created by Iflytek for the conference made a wonderful appearance at the Shanghai World Expo Center and the service desk of the Expo Exhibition Hall. With the blessing of the Spark model, the audience can interact with it more naturally and in real time to understand the trends and hot topics of the artificial intelligence conference. IFLYTEK heard that simultaneous interpretation provided technical support for the opening ceremony and the main forum of the conference, helping the guests from home and abroad to achieve barrier-free communication.
In addition, the Spark Big Model based on domestic software and hardware is also exhibited simultaneously in the AI Big Model Technology Zone and Huawei Shengteng Ecological Exhibition Zone, showing that the Spark Cognitive Big Model empowers diverse ecology of all walks of life, so that users can understand the application results of the Spark Big Model in education, medical care, office, digital employees, automobiles, finance, industry and other scenarios. In front of Iflytek AI learning machine, hearing smart screen, smart office book and other products, the audience rushed to queue up for experience, and the applause continued.
At present, cognitive big model has become the dawn of general artificial intelligence. As one of the national teams of artificial intelligence, Iflytek relies on the long-term solid accumulation, scientific evaluation system and innovation system in the field of cognitive intelligence to promote the continuous evolution of the Spark model, enable the big model to empower thousands of industries and lead the industrial transformation.
Zhongguancun Kejin helps enterprises build "super employees"
Zhongguancun Kejin, as a leading provider of conversational AI technology solutions, made a wonderful appearance at the conference with a new generation of dialogue engines with large domain models, as well as core products and solutions such as intelligent customer service, intelligent sparring, intelligent quality inspection, agent assistant and virtual digital. Among them, the upgraded "Help intelligent cloud Call Center" based on domain model technology attracts many visitors to stop and experience consultation in the form of real-time interaction between voice input and customer service robot dialogue.
The new generation of dialogue engine with help is the basic technical base built by Zhongguancun Kejin based on general big model and domain knowledge base. Relying on the language understanding and content generation ability of domain model, and supported by multi-modal interaction technology, Zhongguancun Kejin has formed a complete technical system, and is committed to building a new production relationship of man-machine collaboration through dialogue AI technology, helping enterprises to create "super employees" and realize strong artificial intelligence applications with analysis and decision-making ability. In terms of technical realization, Zhongguancun Kejin relies on the core competence of domain CoT+ domain big model+domain knowledge base+domain capability suite to ensure the practical application effect. By transforming the domain konw-how into the domain CoT, the big model has more complex business problem solving ability.
For a long time, Zhongguancun Kejin insisted on exploring the innovative application of conversational AI technology in the enterprise service track, closely tracking the development trend of large models, and quickly updating and iterating the technology. Based on the new generation of help dialogue engine, the original products such as intelligent outgoing call, intelligent customer service, intelligent quality inspection and intelligent sparring are upgraded in an all-round way, and virtual employee assistants are introduced to provide enterprises with large-scale model applications that are out of the box, seamlessly connected with the system and affordable, and help enterprises build a new production relationship with high quality, intelligence and temperature.
Softsoft Wisdom released the first white paper on AIGC application in urban governance industry in China.
On July 6th, iSoftStone, a leading provider of urban data intelligence services in China, jointly released the first domestic white paper on the application of generative artificial intelligence (AIGC) in urban governance industry-the White Paper on the Application of AIGC Empowering Urban Governance. At this stage, with the iterative development of deep learning algorithm, generative artificial intelligence technology accelerates the integration with urban digital development, and the field of urban governance has become an important scene for the application of large model technology.
According to reports, from the perspective of urban and industrial development, the white paper deeply analyzes the development status and trend of generative artificial intelligence represented by large model technology, analyzes and combs the innovative application of AIGC to empower urban governance in a scene form, and scientifically explores the path and practice of building intelligent infrastructure in cities under the new situation. The white paper focuses on the Softcom Smart Kongming City Model. Based on the supporting ability of iSoftStone’s intelligent GPT-Creator AI application empowerment platform, it provides prompt engineering tools, domain fine-tuning model base, urban service plugin store, AppBuilder application low-code development platform, pre-training basic multi-model adaptation and other functions for application developers, and enables the rapid construction of AI scene applications.
At the same time, the white paper lists three innovative practice paths and achievements of Softcom Smart Kongming City Model in urban governance, which provides demonstration reference and new ideas for the digital development of cities and industries. Under the opportunity of the development wave of generative artificial intelligence and big model, strengthening social service based on AIGC and coordinating human resources management at the grassroots level will help to further improve the efficiency of social governance in the city and improve the service temperature of people’s livelihood.
Xinghuan Technology Opens a New Journey of GenAI
With the advent of the Universal Big Language Model (LLM), the interaction between human beings and data is undergoing major changes, and the application scenarios of multimodal big data in the financial field are changing with each passing day. The landing practice of LLM is the focus of current industry. Relying on the advantages of big data basic software, Xinghuan Technology exhibited the technologies and products of the whole cycle of big model application and development at the conference.
In the computing power and basic software layer, there are many basic softwares and tools, including star ring distributed vector database Hippo, distributed graph database StellarDB, big data development tool TDS and so on. In the developer tool layer, Sophon LLMOps, a unified operation and management platform for large models, is provided to train and optimize the models. In the application layer, it is a user-oriented domain model, such as infinite Infinity of star ring and SoLar search of star ring.
Among them, the Star Ring Infinity Financial Model is the first large-scale model in the industry for financial intelligent quantitative investment and research, which will play a great role in the fields of financial investment and research, quantitative investment and intelligent reasoning, effectively assist analysts, researchers and investment managers in their daily work, help enterprises better cope with the complex market environment and business needs, and continue to promote cost reduction and efficiency improvement and technological innovation in the whole industry.
Honey constructs a new paradigm of language intelligence
As a language intelligence technology enterprise, Honey not only presented a number of AIGC products to the booth in the form of immersive experience, but also released the first large language model of knowledge quiz and content generation supporting the domestic software and hardware operating environment-Honey Nest on July 6th. The large language model of Honeynest knowledge quiz and content generation is based on hundreds of billions of high-quality Chinese multimodal data training, which can not only realize the customized content generation of "thousands of words and thousands of faces", but also provide more efficient, safe and stable protection for users’ daily use based on Huawei’s rising AI ecology.
Deeply cultivated in the field of language intelligence for more than ten years, we always take artificial intelligence technology as the core, and provide intelligent solutions for various office scenes of governments and enterprises by continuously exploring the innovative application of multi-modal and multi-language intelligent technology. In this sub-field, Honey not only maintains a strong advantage in the precipitation and accumulation of high-quality Chinese corpus, but also has rich experience in insight into user needs. Its self-developed large language model can bring powerful empowerment to government and media users from the perspective of better meeting the needs of the scene.
Honey has always regarded language intelligence as the core development direction, and in the era of big model, it has continuously improved its model ability with the help of hundreds of billions of high-quality Chinese corpus accumulated in the past. By returning to the scene and demand itself, it has accurately solved the pain points in the office field with AI ability, constructed a new paradigm of language intelligence, and served thousands of industries in this innovation.
Yunzhisheng brought a large model of mountains and seas and a series of scene applications to the scene.
At the WAIC conference, Yun Zhisheng brought a large model of Shanhai to the scene, which was the first public display since the release of the model. In the Yunzhisheng booth, the audience in the Shanhai model area is in an endless stream, and everyone is scrambling to ask questions, which leads to an experience upsurge-obtaining information, learning knowledge and seeking inspiration … By inputting relevant instructions, Shanhai model can meet the different needs of different users and strive to become assistants, friends and experts around users.
As Yunzhisheng continues to explore more possibilities for deep integration with specific scenes. At present, Yunzhisheng Shanhai Big Model has penetrated into specific scenes such as smart medical care, smart education, knowledge management and smart marketing, and the scene application based on Shanhai Big Model is constantly enriched and expanded.
The intelligent upgrading of many scenes such as medical care, transportation and the Internet of Things is not only the deep empowerment of Shanhai model technology, but also rooted in Yunzhisheng’s years of industrial practical experience. In the future, Yunzhisheng will accelerate the evolution of general capabilities based on the large model of mountains and seas, and at the same time continuously enhance the landing capability of the industry, expand the application boundary of large model scenarios, and contribute new strength to the intelligent upgrading of thousands of industries.
Lanzhou technology Mencius big model focuses on vertical professional track
Lanzhou Technology made its debut in WAIC with Mencius GPT model, and discussed the trends, opportunities and challenges of artificial intelligence development with the world’s top artificial intelligence experts and industry leaders. At the same time, Dr. Zhou Ming, founder and CEO of Lanzhou Technology, delivered a speech entitled "How Big Models Empower Thousands of Industries" at the "MaaS: A New Model-Centered AI Development Paradigm" forum, deeply analyzing the practice and future development trend of big models and AIGC.
Lanzhou Technology booth focuses on Mencius GPT model, focusing on vertical fields and professional racetracks, and focusing on the realization of responsible, safe, controllable and more powerful general model (L1), industry model (L2) and function engine (L3). The L1-L2-L3 three-level service system for B-terminal landing has attracted a lot of attention and consultation from the audience and the media. At the same time, at the invitation of the organizer of the artificial intelligence conference, Lanzhou Technology also showed the latest achievements and progress of Mencius big model technology in the special theme exhibition area "Towards General Artificial Intelligence".
In the era of global intensive and active innovation, NLP technology has entered the stage of industrialization and implementation. Artificial intelligence represented by the big model is promoting a new round of scientific and technological revolution and industrial transformation to accelerate the evolution, bringing about the leap-forward development of cognitive intelligence technology. Lanzhou Technology proposed a three-level service system of Mencius Big Model "L1-L2-L3". Based on the general big model, it added industry knowledge to build an industry model, and then based on the industry model, it integrated common tasks, from the general model to the industry model, and then to the specific scene model of the enterprise, thus further providing better enterprise services.
At present, the large-scale model has become the "highland" of global scientific and technological competition. The large-scale model has gradually become more subdivided, more vertical and more professional from its initial spelling power and scale. With the arrival of the new technology era, entity enterprises can’t wait to embrace the big model technology, promote the intelligent upgrade of design, sales and service, and improve the efficiency of production, operation and management. Whether or not to have a big model will become an important watershed between technology giants and other technology platform enterprises. In the future, the basic big model may present a fierce competition pattern. As the focus of the industry’s eyes, it will eventually become a new generation of traffic portals.
Text/Jin Renfu
Editor/Fan Hui