There is huge room for imagination in large-scale artificial intelligence models. “AI ferryers” in the financial field have emerged.

Recently, Wu Zhaohui, Vice Minister of the Ministry of Science and Technology, pointed out at the 2023 Zhongguancun Forum that the ChatGPT large model technology has kicked off the move towards general artificial intelligence, will trigger changes in the economy, society, culture and other fields, and is expected to promote human society to move towards intelligence. The era of augmentation.

"No one thought that after obtaining enough huge data, the large model would emerge with such amazing capabilities. Compared with the human brain, the neuron connection weights of the large model are far from reaching the level of the human brain, but the human brain Evolution has taken hundreds of millions of years, and the evolution speed of large models will far exceed our imagination." Li Houqiang, deputy dean, professor and doctoral supervisor of the School of Information Science and Technology, University of Science and Technology of China, said at the finals of the "Tianma Cup" National College Science and Technology Competition .

Since the emergence of ChatGPT, the imagination space of artificial intelligence has been infinitely expanded. Experts have high hopes for large artificial intelligence models. Major companies are also eager to try and increase their layout and exploration efforts. AI large models may be becoming a thing not to be missed. historical opportunity.

Financial industry needs “AI ferryman”

In fact, every climax of artificial intelligence is promoting the rise of artificial intelligence. If we are now in a huge software Internet ecosystem, then in the future we may enter a huge artificial intelligence Internet ecosystem. The capabilities displayed by the current large AI models are enough for people to imagine future technologies. blueprint.

No one can predict the direction of technological development, just like no one can predict the outbreak of ChatGPT, but someone has to climb the "Mount Everest" of technology.

The potential of AI large models is obvious to all, and there are countless new players who have joined AI large models. Various fields are constantly exploring the unknown areas of large AI models through innovative forms, either increasing investment in scientific research, visiting companies to learn from experience, or holding technology competitions to brainstorm ideas.

The "Tianma Cup" National College Innovation Competition mentioned above is the fourth technology-themed competition hosted by Immediate Consumption, a technology-driven financial institution holding a consumer finance license, and the theme of this year is "Yuanqi" "Infinite, AI Creates the Future" is an "exploration journey" of large AI models to jointly find effective paths for the implementation of cutting-edge technologies such as generative AI and NLP. It is reported that the competition has attracted active students from 110 universities at home and abroad. participate.

It can be seen that large model technology is indeed attractive enough, and will indeed have a certain impact on human life, production, economy and society.

Wu Zhaohui also said that it is necessary to continue to strengthen the innovation and development of large-scale model technology, promote the implementation of industry applications, strengthen ethical risk management, and allow artificial intelligence to better benefit human society.

But in fact, AI large models will still face several problems in their development. Li Houqiang summarized it as: the problem of AI chip stuck neck, lack of technical details in the research and development process, and difficulty in obtaining data.

In fact, not only the underlying large model technology needs to evolve, but in the face of a new round of technological iterations, the financial industry as the underlying support must also make corresponding changes with the advent of the artificial intelligence era. "AI large models will bring huge changes to the financial industry," Li Houqiang said.

As the competition organizer, we will leverage our expertise in consumer finance as soon as possible and use reinforcement learning from human feedback to train language models. Judging from the current practical results, its highly anthropomorphic virtual customer service can handle repeated problems simply and quickly, reaching an accuracy of 95%, reducing financial service costs and improving user experience; relying on intelligent network security, it has deepened the relationship map of 74 million users and provided intelligent The robot has served more than 1 billion times, and the full-scenario secure verification of trusted face recognition has exceeded 200 million times.

Artificial intelligence technology is running through the financial industry chain, injecting strong impetus into the high-quality development of the financial industry. Immediate consumption plays the role of "AI ferry" in the financial industry.

Collaborate with "industry, academia and research" to solve the problem of large model industry implementation

It is important to emphasize that due to the lack of scientific and technological talents, the core problem in the current development of large-scale artificial intelligence models lies in talent training and talent introduction in universities.

It can be seen from the "AIGC Talent Trend Report" that since the launch of Open AI in 2021, the competition for AIGC-related talents has begun. From January to February 2021, AIGC-related job recruitment increased by 281.88% year-on-year. Then in 2022 and January-February this year, the number of recruitments maintained a year-on-year increase of 76.74% and 31.3% respectively.

Of course, the cultivation of talents cannot be achieved overnight. It requires a long training cycle and allows talents to smoothly transition from the "academic ivory tower" to the actual industry. In addition to considering the advanced development of artificial intelligence technology, it is also necessary to solve the problem of combining AI with professional knowledge in specific industries. Only when it is consistent with the attributes of the industry can artificial intelligence achieve true technology promotion, thereby bringing actual value to enterprises or industries in terms of cost reduction and efficiency increase. The "Tianma Cup" competition is an event that puts the theoretical knowledge of college students into practice.

Technology can only exert its value when it is implemented and applied, and theory must be combined with practice. Li Houqiang said in the interview, “Because technology not only requires cutting-edge research, but more importantly, it must be implemented and applied, and then it can develop automatically.”

consuming immediately is well aware of this, and has jointly proposed the task of multi-party empathic dialogue with Nankai University, and invented a dialogue technology with empathy for multi-person communication, which can make robots more empathetic He carefully participated in multi-person conversations, and the paper he wrote based on this was accepted by ACL 2022; he jointly applied for the "Research and Industrialization of Key Technologies for Intelligent Data Governance and Privacy Security in Cloud Platform Environment" project with Sichuan University and others and also won the Wu Wenjun Artificial Intelligence Science and Technology Award. Awards etc.

The latest news disclosed that MaMaiConsumption has jointly built laboratories with 16 universities and research institutes including the Chinese Academy of Sciences, Nankai University, and Huazhong University of Science and Technology, and has conducted research on face recognition, knowledge graphs, deep learning, customer service robots, and dialogue text. Breakthrough progress has been made in abstracts and other projects, and the transformation of results has been accelerated. Up to now, more than ten top conference papers have been published, including ACM, MM, EMNLP, ACL and other journals. The journal sources are CCF-A, CCF-B, etc.

Li Houqiang said that the domestic research and development of large AI models is progressing rapidly, and more possibilities will be born in the future.

It should be noted that foreign countries have begun to try to formulate industry standards, and behind industry standards, it mainly involves technical safety. Because the more powerful the technology, the higher the risk, especially in vertical industries. The financial industry is the most sensitive. Domestic financial institutions are now actively promoting the formulation of relevant standards, focusing on two aspects: normative constraints and risk monitoring.

Artificial intelligence technology is accelerating its deep integration with the financial industry. With the development and continuous optimization of technology, it will bring changes to the financial industry and drive financial services to be more efficient, convenient and warm. With the continuous exploration of artificial intelligence large model technology, the financial industry and other industries are expected to usher in an era of comprehensive intelligence.

The above is the detailed content of There is huge room for imagination in large-scale artificial intelligence models. “AI ferryers” in the financial field have emerged.. For more information, please follow other related articles on the PHP Chinese website!