After learning how to build and use AI intelligent agents, self media operator Xiaomi feels like it has hired an extra "copywriter" who works 24 hours a day for itself, and can generate 10 popular headlines in just one minute, "much more worry free". AI agents are becoming a new buzzword in artificial intelligence. As a promising direction for the development of AI applications, AI intelligent agents have attracted much attention. Not long ago, OpenAI listed a five level roadmap for implementing general artificial intelligence: L1 is a chatbot; L2 is an AI that can solve problems just like humans; L3 is an intelligent system that can not only think but also take action; L4 is an innovator; L5 is the organizer. What exactly is an AI intelligent agent? Simply put, an AI agent is like a 'little assistant' with intelligence, emotional intelligence, understanding, and assistance Chen Hao, Deputy Director of the Advanced Technology Center at Beijing General Artificial Intelligence Research Institute, said. What is the difference between AI agents and large models? Pu Zhiqiang, a researcher at the Institute of Automation of the Chinese Academy of Sciences, said that AI agents can be understood as a more three-dimensional, "humanoid" intelligent system. In addition to providing widely used language communication for large models, AI agents can also complete more complex tasks. For example, when the task instruction of "buying coffee" is given, the big model will tell you "I cannot directly buy coffee for you" and give other suggestions; The AI agent will first break down how to purchase coffee and formulate steps to place orders and make payments on behalf of a certain app. Following these steps, the app will be called to select takeout, and then the payment program will be called to place orders and make payments, without the need for the person giving instructions to specify each step. At present, AI intelligent agents have been applied in many scenarios, such as customer service, programming, content creation, knowledge acquisition, finance, investment research, mobile assistants, industrial manufacturing, etc. On Baidu ERNIE Bot APP, everyone can create and display their own AI agents, carry out "video dialogue", memorize words, correct spoken English, and also through AI agent simulation interviews, conversations with AI ancients, etc. Scoring and optimizing resumes can be done in just a few seconds; Generate various styles of copywriting; After teaching the job content to AI agents, they can help themselves complete a part of the work, which is equivalent to having their own interns... In January of this year, the personalized intelligent agent customization function of Zhipu was launched, and users can create their own AI agents with simple prompt words and instructions. According to the person in charge of Zhipu, the intelligent agent center has hundreds of thousands of users who publicly release intelligent agents, with an average of over 500000 questions per day. With the rapid development of AI technology, many people are using it more as a "tool" rather than a "toy" to improve work efficiency in daily life. Xiaomi, a self media operator with 10 years of brand operation experience, has helped many small and medium-sized enterprises build their own AI intelligent agents. With its help, one person can quickly produce 300 pieces of copy and manage multiple accounts at the same time. More and more AI agents are being applied in various industries, which will bring about productivity improvements and a quiet transformation in enterprise operation and management methods. Pu Zhiqiang noticed that the enthusiasm for institutional deployment of "digital employees" is high, with rich application scenarios and increasingly complex tasks. Some segmented markets have undergone disruptive changes. In the content creation industry, AI agents have been able to generate videos from a single sentence or use assembly lines to collaborate with multiple large models to produce more complex short films; In literary creation, the use of AI intelligent agents to generate scripts, ideas, etc. has been accepted by more practitioners. The widespread application of AI agents can reduce labor costs. For ordinary people, it is equivalent to having many helpers; For enterprises, it is also an opportunity for intelligent transformation Lanzhou Technology CEO Zhou Ming said. Chen Hao said that the general AI intelligent body can actively participate in various aspects, rather than passively executing preset tasks. This not only helps improve production efficiency, but also better responds to unexpected situations and enhances the flexibility of production and services. In addition, AI agents also have a more positive role in solving social problems such as aging. According to Chen Hao, the "embodied intelligent robots" and "high-density tactile dexterous hands" of Beijing General Artificial Intelligence Research Institute can recognize the environment, adjust action strategies according to the environment, and do not require additional instructions from humans. AI intelligent physical fitness assistance will be a great helper for elderly people's daily life and home-based elderly care. In addition, in places where people cannot reach, such as major natural disaster sites, AI intelligent agents can be used to carry out dangerous operations or rescue tasks, significantly reducing casualties and losses Chen Hao said. At the same time, the development of AI intelligent agents can also open up innovation space for more enterprises and form a more complete AI industry chain. As of August this year, the number of developers on Baidu Wenxin intelligent agent platform has reached 600000, and the number of enterprises has risen to 100000. He Junjie, General Manager of Baidu Mobile Ecology Business Group, stated that AI agents will reshape the way people interact with technology, bringing new application ecosystems, traffic patterns, and business models. How to balance "wisdom" and "governance"? Many industry insiders believe that AI intelligent agents are the future trend. The report '2024 Trends in Digital Technology Frontier Applications' shows that AI agents are expected to become the next generation platform. Robin Lee, the founder of Baidu, also said that he was most optimistic about the development direction of AI applications in the future. Some argue that in the foreseeable future, there will be an AI agent on smartphones that will gather various types of data to meet users' needs anytime, anywhere; It will be the "AI programmer" of enterprises, assisting developers in completing a series of tedious tasks; Or a 'data analyst' who automatically collects and analyzes enterprise data, mining the value of data gold mines. Robin Lee said that with the growing strength of the basic model, development and application are becoming easier and easier. As long as the workflow is explained clearly with "human words", coupled with a proprietary knowledge base, a valuable AI agent can be made, which is simpler than making a web page in the Internet era. Many interviewed experts also believe that the current AI intelligent agents can only complete relatively simple and fixed tasks, and are still in the early stages of development overall. At the same time as rapid development, AI agents also face technical risks, ethical and privacy issues. Industry insiders say that AI agents collect a large amount of data during the process of providing services, and it is necessary to ensure data security and prevent privacy breaches. Although AI agents can provide suggestions, not all of them have authority. In addition, the stronger the autonomy of AI agents, the more likely they are to make unpredictable or inappropriate decisions in complex or unforeseen situations. Zhou Ming believes that security barriers should be established from different dimensions: in the research and development process, security should be taken as the core design principle, and corresponding training should be provided to R&D personnel to prevent model algorithms from being biased; In the process of use, enhance public awareness of the boundaries, risks, and challenges of technological applications, and increase feedback channels. The relevant person in charge of Zhipu stated that enterprises should establish strict internal standards and operating procedures to ensure that development, testing, deployment, and use are all within a controllable range. Promote the formulation of industry standards and provide guidance for the healthy development of AI intelligent agents. At the same time, a responsibility traceability mechanism should be established. Zhang Haifeng, associate researcher of the Institute of Automation of the Chinese Academy of Sciences, also believes that policy formulation should incorporate the views of multiple stakeholders, establish an appropriate AI ethical review mechanism, and conduct independent review of high-risk projects to ensure that they meet ethical standards and are beneficial to society. (New Society)