Strategy Ai Xiongfeng | Which AI applications can be implemented quickly--TMT Rise Series (4)

Jinxuan·Core Viewpoint

1. How to compare the implementation progress of AI applications in different fields: the framework of investment and fault tolerance

AI application fields can be divided into three levels from easy to difficult in terms of technical difficulty: "helping decision-making, assisting creation and replacing execution". 1) Decision-making assistance is when AI forms knowledge based on data and information, and then helps humans make decisions and complete specific tasks that do not require high accuracy. Mainly used in life, office and professional services. For example: intelligent assistant: daily life, office management, etc.; professional services: advertising, education, finance, medical care, logistics, security, electricity, etc.; 2) Assisted creation is the formation of logical reasoning ability by AI based on knowledge to assist content creation. Achieve creative goals. Mainly used in information, text, images, movies, games, etc.; 3) Substitute execution is the formation of high-precision execution capabilities of AI based on logical reasoning. It is mainly used in the field of intelligent machines to replace humans in executing solutions that require high precision. Such as smart cars, smart robots, smart factories, etc.

The implementation of AI applications depends on the investment of companies in the application field on the one hand (including R&D investment and capital expenditure, etc.), and on the other hand it also depends on the error tolerance rate of the application field (generally speaking, the trial and error cost is low AI application fields are often easier to implement). From a bottom-up perspective, application areas with high investment and high fault tolerance tend to be implemented faster. In some fields, despite heavy investment, the fault tolerance rate is low, and the implementation of AI applications in these fields generally takes longer. In addition, for those fields with low investment, if the cost of trial and error is high, it will be difficult to implement the application. Some industries may have higher fault tolerance rates, but low investment often means less room for application implementation.

2. Computers: Office suites, financial IT, enterprise services and other industries are expected to be the first to be implemented

Computer application scenarios are rich. Although the proportion of C-side is not high, it still includes office software, securities IT, intelligent hardware, etc.; B-side scenarios are more abundant, and the demand is relatively market-oriented, including industrial software, enterprise services, Financial technology, etc.

Office Suite: The current wave of AIGC technology driven by large models has an impact in the field of office suites mainly in two major dimensions: First, through intelligent identification, analysis, and review of unstructured documents, It greatly improves the efficiency of programmed work; secondly, it provides creative workers with a large number of optional creative materials and becomes an intelligent assistant for creative workers to assist them in their creation.

Financial IT: Generative big language models can empower scenarios such as intelligent customer service, product recommendation, market analysis, risk control, and report generation in the financial industry, helping banks, securities firms and other financial institutions improve service quality. and work efficiency.

Enterprise Services: As an enabler of digitalization and intelligence in all downstream industries, it is expected to benefit significantly. With the launch of large models by major technology manufacturers: Baidu's "Wenxin Qianfan" provides customers with enterprise-level large language model services; Alibaba launches the "Tongyi Qianwen Partner Program" covering various industries, and AGI general capability segmentation scenario model training is expected to be Reshape the puzzle at ERP, CRM, OA, HR and other levels.

3. Media: Games, marketing and other industries are expected to take the lead

Judging from the implementation trend, we believe that the industry represented by "AI content" will be implemented as soon as possible and receive the dividends of this round of AI earlier. The reasons are: 1) The AI ​​ecosystem is prosperous, and all leading manufacturers have released self-developed large models. For content companies, AI capabilities can be directly obtained through direct calls or B-side cooperation. 2) Current AI technology can already help realize simple content creation, and industry technology has preliminary application capabilities.

Game: The implementation scenario of AI game can be divided into two major levels: 1) Cost reduction and efficiency increase in the research and development process: AI can effectively reduce costs in the game production process by virtue of its high efficiency and low cost. This increases efficiency. AIGC technology has shown obvious potential in cost reduction and efficiency improvement in fields such as 2D art batch image generation, basic code review, and AI voice application. 2) User experience upgrade: Intelligent AIBot can bring a stronger sense of interaction as an assistant and NPC during the game, using AIGC to enrich game levels and improve game user playability, etc.

Marketing: Also beginning to take shape, AI advertising will empower content understanding and advertising delivery models. For example: Sanrenxing and iFlytek collaborated to jointly develop the next generation of AI multi-modal intelligent marketing tools. Tencent's advertising end is also connected to the Hunyuan large model and the advertising precision ranking model to optimize and increase efficiency in the entire link from advertising production to push.

risk warning

Economic downturn exceeds expectations, macro liquidity shrinkage risk, overseas black swan events, industrial policy implementation falls short of expectations

The above is the detailed content of Strategy Ai Xiongfeng | Which AI applications can be implemented quickly--TMT Rise Series (4). For more information, please follow other related articles on the PHP Chinese website!