Analysys: Although the development of the artificial intelligence market in 2022 is less active than expected, 2022 will undoubtedly be an exciting year for the artificial intelligence industry. In the middle of the year, text-image generation models such as DALL-E 2 and later Stable Diffusion and Midjourney attracted a lot of public attention to the content generated by artificial intelligence. At the end of the year, the emergence of ChatGPT refreshed the public's perception of the intelligence level of artificial intelligence. .
A series of phenomenal events indicates that the artificial intelligence industry is undergoing profound changes. The progress of the industry has narrowed the long-standing gap between our expectations for artificial intelligence and its functions, and at the same time has further expanded our imagination of artificial intelligence. , Laying a solid foundation for more intelligent applications to enter our production and life. We believe that the reform will eventually clear the fog of development, remove the ice of technology, and make artificial intelligence truly the pillar of the new chapter of industrial development.
Analysys releases a report on the development trend of the artificial intelligence industry in 2023 through cutting-edge applications in the industry, research progress in academia, and investment and financing market conditions, combined with expert opinions, industry views, phenomena-level events, and research accumulation on the artificial intelligence industry.
This report analyzes the key elements of various trends from the three aspects of artificial intelligence industry infrastructure, algorithm models, and industrial applications, and deeply analyzes the development nodes and core driving forces of trends. Analysys believes that artificial intelligence will accelerate its entry into thousands of industries, deeply participate in the construction of digital China and become one of the core driving forces for the high-end industry, and the artificial intelligence industry structure will also usher in rapid changes.
1 Overview
The Analysys artificial intelligence AMC model shows that image classification and image semantic segmentation applications are relatively mature and have a relatively stable market space, and applications such as text processing, speech recognition, and dual-modality are gradually realizing market penetration. Related applications such as reinforcement learning, causal learning, and large language models have successfully stepped out of the laboratory through technology iterations, and are constantly exploring their business models. Applications such as graph neural network, multimodal generalization, and self-supervised learning are speeding up the difficulties from experimental research and development to industrial implementation. Research on diffusion models, quantum AI, and embodied intelligence will also breed higher intelligence and versatility. Stronger applications. It is recommended to pay attention to the application maturity between the market start-up period and the high-speed development period in the short term, and pay attention to the application research and development progress in the exploration period and market start-up period in the long term.
2. Infrastructure
Trend 1
The demand for artificial intelligence development will rapidly increase the scale and professionalism of the data crowdsourcing industry
Analysys believes that the overall development of the artificial intelligence industry has led to a rapid increase in demand for data, which will stimulate the vigorous development of the data crowdsourcing industry. Require. At present, industries such as manufacturing, medical care, and engineering construction have a large demand for functional upgrades of intelligent applications, so the data crowdsourcing industry will be promoted to take the lead in the development of related industries.
trend 2
my country will form a chip-artificial intelligence industry internal cycle
my country's industrial ecology dominated by leading enterprises and core scientific research institutions is relatively prosperous and resilient, which is conducive to the rapid development and iteration of the chip industry. At present, the growth rate of my country's artificial intelligence industry ranks among the top in the world. Based on the huge demand for chips in the artificial intelligence industry, Analysys predicts that it will drive the rapid growth of domestic chip demand. Correspondingly, the application of artificial intelligence is indispensable in circuit design and defect detection, and is of great significance to chip process improvement, production process control and other links. Therefore, Analysys believes that my country will form a "chip-artificial intelligence" industry cycle.
trend 3
Accelerating the exploration of edge intelligence requires close cooperation between different types of participants
Analysys believes that the development of edge computing power has a positive impact on the artificial intelligence industry, but the development and deployment of artificial intelligence applications on the edge side still needs to face the problems of adaptation and optimization. Due to the diverse needs of the industry for intelligent applications and the engineering needs of artificial intelligence application development, it is necessary for manufacturers from different industries and with different core technologies to form a closer industrial ecology to accelerate the process of exploration. In addition, if you want to make better use of the idle computing power brought by the edge computing power, you need to cooperate more closely with the artificial intelligence industry and the computer, communication and other academic circles to form a closed loop of research-development-testing-application.
3. Algorithm Model
trend 4
The text-image generation model will have customized products for the needs of subdivided fields
Analysys believes that to build a sustainable business model for text-image generation applications, we should actively explore the mechanism of sharing profits with upstream galleries. In the fields of industrial design and product design, there are a large number of subdivision needs with obvious professional attributes. Therefore, text-image applications for subdivision needs will appear within two years, and customized development may also become the mainstream business in the next stage of the industry. model.
Trend 5
The direction of commercialization of large-scale language models in professional fields still needs to be continuously explored
Analysys believes that in the next two years, due to the requirements of search and recommendation for knowledge in general and many professional fields, the business model of large-scale language model application in text content production and programming may mature earlier than search and recommendation. Large-scale language models already have the functional basis for application in multiple fields in terms of precise and fuzzy mapping of concepts, logical relationships between concepts, and conceptual reasoning. However, the direction of commercialization of large-scale language models in the next 3 to 5 years is still Continue to explore. Analysys believes that it will take at least 5 years to explore the optimal path for the relationship between large-scale language model training data and parameter levels, and the optimal path can better balance the functions and costs of large language models, which is conducive to the development of large language models. commercial exploration.
Trend 6
Intensive learning applications may be the first to be commercialized in the fields of scientific research and industrial R&D
Analysys believes that ChatGPT's application of reinforcement learning has brought a new perspective to its commercial application, that is, starting from the demand for capabilities, exploring the application direction of reinforcement learning. Based on the strong demand of scientific research and industrial R&D for the capabilities of reinforcement learning in terms of rules, strategies, games and human-like decision-making, Analysys believes that the commercial application of reinforcement learning will turn to relevant markets. Although intensive learning for scientific research and industrial R&D has higher professional requirements, it will be commercialized within three years due to its higher application value.
Trend 7
The commercial value of various applications of graph neural network will be greatly improved
Analysys believes that due to the substantial improvement in functions, the graph neural network will accelerate application penetration. In terms of graph learning, the general model-task matching evaluation method and the more expressive deep graph network (DGN) can improve the application effects of marketing crowd matching, three-dimensional space classification and segmentation, public opinion influence prediction and compound task design, etc. . In terms of graph generation, the application effects of graph recurrent neural network (GraphRNN) and graph convolution policy network (GCPN) in building knowledge graphs and molecular structure discovery have improved significantly. Since there is a great possibility of a qualitative leap in terms of functions, Analysys expects that the related applications of the graph neural network will gradually promote the commercialization process.
Trend 8
Diffusion models will be applied to design, architecture, advertising and other industries within this year
As the current mainstream underlying model for text-image generation tasks, Analysys believes that the performance and functions of the diffusion model are expected to improve rapidly. In addition to text-image generation, Analysys believes that applications such as text-speech generation, super-resolution, and image restoration are expected to be commercialized in design, architecture, advertising, film, and medical industries within one year; such as acoustic signal processing, The application of diffusion models such as point cloud completion and generation, time series completion and prediction will gradually mature within 3 years; for the application of diffusion models with larger commercial space such as semantic segmentation and anomaly detection, the cost and efficiency of them can reach a preliminary level after 5 years. Meet the basic requirements of commercialization.
4. Industrial application
Trend 9
There will be more intelligent applications designed in combination with algorithmic model principles in the industry
Analysys believes that intelligent applications designed from the underlying principles of algorithm models in scientific research have a strong reference value for how to apply artificial intelligence in the industry. The potential market space of industrial intelligent applications designed with similar methods is much larger than the current artificial intelligence market size, and the demand for deepening the development of industrial intelligence will promote the development of such applications, but at the same time, a large number of high-level interdisciplinary talents are needed to support its development.
trend 10
The importance of scientific research artificial intelligence as a national strategy will be further enhanced
Analysys believes that the application of artificial intelligence in the field of scientific research has obvious strategic significance for national development and will continue to increase. Due to the foundation and importance of scientific research intelligence in the development strategy, Analysys expects that artificial intelligence-enabled scientific research services will form a considerable market space in three years, but due to the technical threshold brought about by the combination of scientific research services and artificial intelligence, It is extremely difficult to enter the relevant market, so it is necessary to accelerate the exploration of a more reasonable market mechanism.
trend 11
The application penetration rate of smart devices in the industrial field will increase rapidly
At present, the demand for smart devices in the industrial field is growing steadily, while the average unit price of industrial smart devices is gradually decreasing. Analysys predicts that cloud-edge-device collaboration and industrial communication will be popularized in five years, the intelligence level of industrial smart devices will also be greatly improved, and the application value of industrial smart devices will rapidly increase. Therefore, Analysys expects that smart devices in the industrial field will The market size of the market will grow steadily in the next 5 years, and 5 years later will usher in an inflection point of growth.
trend 12
The demand for mobility assistance in the consumer sector may promote the development of related smart devices
Analysys expects that after five years, due to the gradual maturity of technologies such as motion posture estimation and additive manufacturing, the cost of mobility aids such as prosthetics, sports rehabilitation equipment, and aging equipment will gradually decrease. There will also be a significant increase in related demand, and the acceptance of high customer unit prices in related fields is relatively high. Therefore, compared with mobile smart devices in other consumer fields, the commercialization of mobility-assisted smart devices will develop first.
Disclaimer: The third-party data and other information quoted by Analysys in this article are all from public channels, and Analysys does not assume any responsibility for this. In any case, this article is for reference only, not as any basis. The copyright of this article belongs to the publisher. Without the authorization of Analysys Analysis, it is strictly forbidden to reproduce, quote or use any content released by Analysys Analysis. Any media, website or personal use after authorization should quote the original text and indicate the source, and the analysis point of view is subject to the official content released by Analysys Analysis, and any form of deletion, addition, splicing, deduction, distortion, etc. . Analysys does not assume any responsibility for disputes arising from improper use, and reserves the right to pursue responsibility against the relevant responsible parties.