Article source: [Length Youshu Technology]
Since the release of ChatGPT, the popularity of AI algorithm models has continued to heat up. Experts disclosed that so far, hundreds of large-scale models with a parameter scale of more than 1 billion have been released in China. The World Artificial Intelligence Conference also released a batch of large-scale models in July, and a "hundred-model war" officially started.
Attached "Research Report on China's Artificial Intelligence Large-scale Model Map" released some representatives of domestic large-scale models
However, Emad Mostaque, CEO of artificial intelligence (AI) startup Stability AI, said in a conference call recently that AI is a $1 trillion investment opportunity, but it will also be the biggest bubble ever.
So, what is the nature of this AI technology revolution? Is AI an outlet or a bubble?
"Hundred Models War" maybe just a bluff?
It is understood that domestic and foreign technology giants such as Microsoft, Google, Baidu, and Ali have responded quickly and successively launched various AI large models to seize the market.
However, when talking about the "100-model war" in online public opinion, Zhang Dixuan, President of Huawei's Ascend Computing Business, said in an interview with "China Business Daily" that large-scale models have become a development trend, pushing AI towards general artificial intelligence. At the Xinli meeting of Meihua Venture Capital, Zhu Xiaohu, managing partner of GSR Venture Capital Fund, believed that large models had no investment value.
In this regard, Dr. Yuan Yue, chairman of Zero Point, pointed out in an exclusive interview with Yiou that the "100-model war" may be just a bluff. "Although the current so-called large models have attracted widespread attention and give people the impression of a large number, most of these models have not really entered the actual combat scene in the actual application and competition fields of users, nor have they shown actual competitiveness." he said.
Yuan Yue pointed out that compared with international ones, my country's algorithm model lacks competitiveness in the entire large-scale model project. In fact, the number of Chinese models that can really reach the level of large models is limited, and there is no model that can rival ChatGPT. No matter in terms of investment preparation, development goals, or application experience, there is still no solution corresponding to GPT technology.
Attached picture Yuan Yue shared the theme at the 3rd BPAA Algorithm Application Practice Model
The current algorithm model construction is still difficult?
The industry believes that technological breakthroughs represented by large models have significantly lowered the threshold for artificial intelligence to empower thousands of industries, indicating that artificial intelligence technology will be widely used in various fields and promote industrial upgrading and transformation.
However, the current algorithm model construction is facing many challenges, and there is a long way to go. It is understood that at present, enterprises often face problems such as low data quality and insufficient scene opening when applying large models. On the other hand, artificial intelligence projects, smart city construction, and smart industry construction in various places have insufficient budgets for model development, and there have been no major adjustments in the concept and structure of model development investment among partners, customers, and suppliers.
In addition, the era of large models also puts forward higher requirements for computing power. Wanka-level computing power is a prerequisite for training large models and an inevitable requirement for promoting the industrialization of algorithms. This requires more investment in algorithm ecological construction intensity. When talking about the next step of the current artificial intelligence industry algorithm ecological construction, in an exclusive interview with the Organizing Committee of the 3rd BPAA Applied Algorithm Practice Model, Secretary General of Shanghai Artificial Intelligence Industry Association and Secretary General of Shanghai Artificial Intelligence Standardization Technical Committee Zhong Junhao said that he hopes that more social capital can enter the construction of algorithmic ecology, but don't follow GPT hotspots in a rush, but invest steadily with the progress of science and technology.
Attached picture Zhong Junhao, Secretary-General of Shanghai Artificial Intelligence Industry Association and Secretary-General of Shanghai Artificial Intelligence Standardization Technical Committee, was interviewed
Where will the future of "Hundred Models Battle" go?
According to Zhou Hongyi, the founder of 360 Group, large models need to be versatile in order to empower thousands of industries. Large-scale models will be diversified in the future, and the development trend is to be small and specialized, and the real opportunity lies in the enterprise market. He mentioned at the Tianjin GPT Industry Development Plan Launch Conference that, in fact, although leading companies continue to launch large-scale models one after another, how to make full use of the existing large-scale model foundation, give full play to the advantages of the scene, and form an industrial ecology of general artificial intelligence. All industries are still exploring.
Similarly, Yuan Yue, chairman of Zero Point Youshu, said in an exclusive interview with Yiou that it is not only necessary to invest in the field of algorithm models, but also to strengthen investment in vertical mechanisms, intelligent construction and content construction. On the other hand, the lag of the existing talent education mechanism requires the algorithm model industry to actively participate in the frontline of algorithm model talent training, providing full support from system concepts, development skills, to application methods, and actual case training.
In addition, the creation of a better platform and environment supported by the government, the gathering of upstream and downstream partners in the industry chain, and the promotion of cooperation among multiple parties such as technology, resources, and scenarios will help promote the development and application of large models. In an interview with the People's Daily, Jiang Guangzhi, secretary of the party group and director of the Beijing Municipal Bureau of Economy and Information Technology, put forward three countermeasures and suggestions based on this: First, strengthen policy innovation and standard guidance, and support small and medium-sized enterprises based on landing application scenarios to obtain diversified, low-cost The second is to increase the opening of scenarios, take the lead in implementing applications in government services, smart cities and other fields, and improve the modernization level of urban governance capabilities; the third is to accelerate the implementation of large-scale industry applications, and further develop the marketization of partnership programs Integrate the role of the cooperation platform to create a good ecology of large-scale model innovation cooperation and application landing.