AI Leads E-commerce and Live Streaming in China

Ever had this experience? You open TikTok, just intending to check out some fun short videos, and accidentally find yourself in a live stream. Suddenly, you’re captivated by the engaging pitch of the host, so you pause to see what’s going on. For the next hour, it’s like you’re hopping between various live streams, effortlessly swiping up and down, with different products popping up before your eyes. With the incredibly good deals and the host’s enthusiastic recommendations, you might order a few items.

Yep, that’s the hottest shopping trend in China right now — live streaming sales.

What is Live-streaming sales?

Live streaming sales is a new sales method that integrates “live streaming” with “online shopping”. Nowadays, online shopping has become one of the main ways for Chinese people to shop. In the past, the traditional form of online shopping mainly involved merchants displaying products on various platforms for users to choose from, with promotional materials primarily consisting of images and text. In 2016, China’s largest shopping platform, Taobao, launched live streaming functionality, and subsequently, a group of live streamers gradually emerged. By 2018, short video platforms like Douyin (TikTok) began to develop live-streaming e-commerce businesses, combining short video platforms with sales, leading to another surge in online shopping activity. At the same time, the group of live streamers began to become more diversified, with any business or individual being able to complete the process of promoting products simply by using a smartphone to shoot videos. In 2023, China’s short video user base reached 1.06 billion people, accounting for 95.1% of the internet user base, with 530 million users participating in online live shopping. The size of China’s live streaming e-commerce market reached 4.9 trillion yuan and continues to grow. (iResearch Center, 2024)

AI live streaming knows me so well!

What is AI

In recent years, the term AI has gradually entered the public consciousness. When we mention AI, we may think of our smartphone assistants, car navigations, or AI-generated art and writing. To understand AI, two words cannot be avoided, as they together drive the development of AI: algorithms and data.”

Algorithms are like the soul of AI, defined as the rules and processes established for activities such as calculation data processing, and automated reasoning (Flew, 2024). They provide AI with the ability to create and think, enabling it to complete tasks autonomously.

Data is the fuel of AI. In the field of AI, it primarily refers to the datasets used for training and testing AI models. Algorithms can be improved over time by learning from repeated interactions with users and data how to respond more adequately to their inputs (Flew, 2021). AI systems are not autonomous, rational, or able to discern anything without extensive, computationally intensive training with large datasets or predefined rules and rewards (Crawford, 2021). Therefore, Big data and algorithmic selection are co-evolving (Just and Latzer, 2016).

AI has rapidly transformed people’s lifestyles and modes of production. It has found wide-ranging applications across various domains including urban development, social management, and national security. AI technology has also become a new force in China’s rapidly growing live streaming e-commerce industry. The content recommendations we receive are results derived by algorithms based on our information. The videos, texts, and sounds we see are all generated and processed by AI. Can you even imagine when you click on a live stream, a host is vividly introducing products, it is not a human being existing in the real world, but a completely computer-generated digital persona?

The advantages of applying AI technology in live streaming e-commerce are evident: thanks to powerful algorithms, our online shopping experience feels more like a “personalized service”. Unlike traditional television sales, where manufacturers typically use generic scripts to sell products, personalization happens on the basis of one’s own user characteristics and own user behavior, others’ user behavior, information on user-connectedness, and location (Just and Latzer, 2016). This means when we browse the internet, computers are also capturing our inputs and selections to form a database for us individually.

In addition to data processing, AI also brings more possibilities to live streaming. Let’s look at the following example, which demonstrates the application of AI technology in the visual aspect of live-streaming. At the same time, think about a question: Do we have the ability to distinguish AI, and is the use of AI always “feasible”? In such a vast market, how should we govern AI live streaming?

AI digital human

In the corners of urban streets, the lights of 24-hour convenience stores shine brightly, even in the wee hours. Similarly, a few uninterrupted live streams appear in late-night shopping apps. After observing for a week with suspicion, my friend asked me a question: “Why every time I open these live streams, it’s always the same hosts? Don’t they ever take a break?” My answer shocked her – these lively hosts are not real humans at all; they are entirely digital humans generated by AI technology. Unlike virtual hosts, there is no “operator” behind them. From appearance to movement to voice, everything about these digital humans is generated by AI. They can engage in intelligent interactions with humans and even possess thoughts and emotions.

Last year, Du Hua, the CEO of one of China’s largest entertainment companies, Yuehua Entertainment, created a digital human named Hua Hua Zi based on her own image and applied it to live-streaming e-commerce. Her livestream attracted 300,000 viewers within just ten minutes of going live, without any promotion. She later created digital personas for the company’s artists, who are also expected to enter the field of live-streaming e-commerce in the future. Clearly, compared to traditional hosts, these celebrities’ faces are more familiar to the public and can also drive fan economies.

(Hua Du and her Digital Figure ”Hua Hua Zi“)

As these leading companies begin to push AI-driven live-streaming sales, it’s becoming a trend for more companies and individual merchants to enter this field. In response to the demand for economic development, China has also invested a significant amount of money to promote the development of AI technology. E-commerce giant Alibaba pledged to spend $15 billion over the next three years to recruit top talent for AI projects (Sun, 2020). However, as a newly emerging technology, AI digital Humans still face many technical and ethical issues.

Does owning a digital human equate to owning money?

Despite tech companies strongly promoting the convenience of AI digital humans to merchants, from an operational standpoint, building a complete, traffic-generating, and profitable live streaming channel from scratch remains extremely challenging. The effectiveness of digital human live streaming does not entirely depend on the maturity of digital human technology; rather, it hinges more on a mature live streaming process. Many companies and individuals face challenges in live streaming standardization even before adopting digital humans. For example, there might not be a stable operating account, and adjustments may be needed for computer and network setups. Without standardized processes throughout the entire live streaming workflow, whether it’s human-hosted or digital human-hosted, it’s difficult to achieve the expected outcomes.

Therefore, there is a growing demand for standardization in AI live streaming. Rather than allowing merchants to figure things out on their own, platforms and governments have a greater responsibility to regulate the entire process of AI live streaming. This is not only for economic benefits but also for a more harmonious market environment.

“Inequitable Transactions” — Privacy Breaches

When we talk about data, a crucial aspect of AI, we turn our attention to a pressing issue: user privacy. In today’s internet society, users exchanging their privacy for platform services seems to have become commonplace. In AI live-streaming, our data is used to analyze our user preferences, enabling better personalization. Despite the service usage agreements signed between platforms and users, these agreements are complicated, vague, and legalistic and typically offer the user an all-or-nothing option (Flew, 2021). That means this transaction appears to be not entirely equitable, and users may find it difficult to confirm how much of their information has been leaked.

On July 13, 2023, the Cyberspace Administration of China, along with six other departments, jointly issued the “Interim Measures for the Administration of Generative Artificial Intelligence Services,” which stipulate that non-essential personal information shall not be collected, the illegal retention of user input information and usage records is prohibited, and user personal information shall not be provided to others (the Cyberspace Administration of China et al., 2023).

Can we trust AI?

When browsing shopping websites, the most common question in the query box is often: “Is the product description genuine?” In live streaming rooms, various hosts passionately introduce products and demonstrate their features to the audience, which, to a certain extent, adds authenticity to the products. However, AI live streaming lacks this functionality. In fact, in this specific scenario, AI is tasked with selling products, so in order to boost sales, the information provided by AI may even be misleading. Such misinformation could range from minor financial harm to consumers to potentially causing physical harm from the products. In December 2023, the governance department of Douyin(Chinese version Tiktok) discovered numerous instances of “AIGC fabrications” during routine inspections. In response, the platform imposed severe penalties such as permanent bans and account suspensions.

Digital Humans are not Human

As AI development is still in its early stages, people have both hopes and concerns about AI. Apart from the societal issues brought about by AI applications, AI itself still has many technical limitations. As early as 2021, the concept of digital twins appeared in the live streaming industry. However, due to the limitations of technology at that time, hosts were mostly represented by cartoon avatars. With the development of new AI technologies, AI can now more accurately learn human actions and behaviors, and the generated images are becoming closer to real people. From this perspective, “realistic” and “human-like” have always been important factors in consumer demand and important guiding indicators for the development of AI technology.

Although AI digital personas have developed to the point where they are almost indistinguishable from real humans in appearance, the inherent complexity of humans is difficult to learn and mimic. According to Maslow’s hierarchy of needs, human needs are divided into five levels. From the bottom of the hierarchy upwards, the needs are physiological (food and clothing), safety (job security), love and belongingness needs (friendship), esteem, and self-actualization (Mcleod, 2022). Among these, the fulfillment of the last two levels is largely dependent on interpersonal relationships. AI lacks the ability for emotional analysis; the emotional states produced by AI are based on the analysis of existing emotional data. However, the rapid, intuitive emotional responses between people, as well as various complex emotions, cannot yet be handled and applied by AI. As a result, consumers’ emotional needs in live streaming rooms are still largely unmet by AI.

The Future Path of AI

In summary, will AI digital Humans become the mainstream? The answer is conservative. Although digital humans bring more efficient economic benefits to merchants, consumers may not fully embrace them. However, innovation in AI technology in live streaming will certainly not stop here. It is not merely a patch for the old model but rather the creation of a whole new path.

In the future, the development of AI is likely to lead to a shift in e-commerce from a seller’s market to a buyer’s market. As mentioned earlier, with the emphasis on “personalization,” AI technology will not only serve as a user’s shopping guide but may also become a shopping assistant. For example, AI can assist in price comparison across multiple platforms and help with search queries. Consequently, stricter data protection measures will undoubtedly be adopted by various countries and regions in the future, and consumer privacy awareness will gradually increase. With the development of AI, deeper needs will also be gradually explored, and the management of AI will become more specific and systematic. The changes that AI brings to e-commerce have just begun, and it is only known that these changes will not stop at AI live streaming. The future of e-commerce will be fundamentally transformative.


Crawford, K. (2021). The atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.

Fan, F. (2020, October 20). Artificial intelligence accelerating nation’s industrial upgrade efforts. Chinadaily.Com.Cn.

Flew, T. (2021). Regulating platforms. John Wiley & Sons.

iResearch Center. (2024, February 29). Report on the Live Streaming E-commerce Industry in China in 2023. iResearch.

Just, N., & Latzer, M. (2016). Governance by algorithms: Reality construction by algorithmic selection on the Internet. Media, Culture & Society, 39(2), 238–258.

Mcleod, S. (2022, November 3). Simply psychology. Simply Psychology.

the Cyberspace Administration of China et al. (2023, July 10). Interim Measures for the Administration of Generative Artificial Intelligence Services. The State Council of the People’s Republic of China.

Be the first to comment

Leave a Reply