Source: Whale Selection Society

Image Source: Generated by Wujie AI

In the last month of December, two noteworthy events occurred in the e-commerce field.

The first event that caught the attention of the public was the change in the top e-commerce platform by market value, with Pinduoduo surpassing the industry giant Alibaba. The second event, also of great interest in the e-commerce circle, was the secret submission of documents by the Singaporean fast fashion unicorn Shein, with an expected initial public offering (IPO) in early 2024.

Behind Alibaba being surpassed in market value is the rapid development of e-commerce platforms such as Temu, Shein, and TikTok, which are leveraging overseas expansion. This has prompted action within Alibaba, with Jack Ma stating internally that Alibaba will change, and quickly establishing a "customer-first, AI-first" strategy in an attempt to pivot and catch up with the model-updating overseas e-commerce platforms, positioning itself to win in the future.

Meanwhile, Shein's secret submission of documents may become one of the largest IPOs in recent years, with a valuation of up to $66 billion. What makes Shein worth paying attention to as a mysterious player fully committed to AI-powered e-commerce is whether it is worth learning from for e-commerce players, including Alibaba.

AI-powered e-commerce, which consists of algorithmic customer acquisition, marketing data analysis, and popular product discovery, has seen the most significant efficiency improvement in the area of AI-generated product images. Taking the generation of traditional product images as an example, it typically requires a model, a photographer, and post-production, with the cost of a day's shoot being around 20,000 to 30,000 yuan. However, using AI may only require an operations staff member to utilize AI to generate models or background images, and then insert the product, significantly reducing costs.

Currently, Alibaba has launched Wanxiangtai, JD.com has launched the Linglong platform, and Shein has fully implemented AI product rendering internally. The biggest difference among these may be that the former two focus on generating auxiliary scenes such as backgrounds and models, while Shein focuses on AI generation of core products.

Shein, which is immersed in the "small order fast response," may have just taken the wrong path in the development of AI e-commerce. The increasing number of copyright infringement complaints against AI-generated products has also led to numerous lawsuits against Shein overseas.

Shein's Aggressive AI E-commerce Model

Reviewing Shein's rapid rise, the secret lies in achieving "small order fast response" in the supply chain. Behind these four words is Shein's introduction of 5,000 SKUs per day, compared to the well-known fast-moving consumer goods brand ZARA, which introduces only 20,000 SKUs per year. Shein's terrifying sell-out rate is 98%, with a hit rate of 48%, and its inventory turnover can be reduced to just over 30 days, while ZARA's is 120 days.

Behind this extreme efficiency, there are many secrets worth exploring. Taking Shein's introduction of 6,000 SKUs per day as an example, this is a goal that is almost impossible to achieve with human labor. In reality, behind the fully managed model, these new SKUs are test-produced in quantities of 100, and many are only represented by images, and are then put up for sale at the terminal. If they are confirmed as hit products, the production scale is then expanded.

However, creating 6,000 new product images every day is also a challenging task. Unlike the platform model of Taobao, where merchants are responsible for new product design and promotion, Shein operates on a fully managed model. Faced with a huge workload, Shein has chosen AI e-commerce, which brings significant efficiency improvements, to quickly complete the task.

Shein's AI e-commerce mainly focuses on predicting trend trends and generating e-commerce images. By analyzing trend data, Shein can provide personalized product recommendations for each user, and Shein's design team can use this system to quickly create design samples.

As fashion trends change rapidly, in order to keep up with the trends, Shein uploads thousands of products every day to meet the platform's pursuit of hit products. Therefore, in order to gauge user interests and preferences, Shein will produce many products through real shooting and AI generation.

For example, the hats in the image below, some foreign netizens believe that the designs and models of several hats may have been generated by AI, but they find the hats very beautiful and affordable, and they want to buy one.

In addition to new products, Shein also makes AI adjustments to many hit products, such as creating a red and white short-sleeved shirt similar to Levi's classic. This ensures to a large extent the hit rate of Shein's new products.

At the same time, in order to quickly deliver new products purchased on the platform, Shein's flexible production capacity supports this. Shein has recruited thousands of small and medium-sized fabric and accessory suppliers, creating an online B2B supplier platform. When a product on the platform is purchased on a large scale, the orders for this product will be sent to the manufacturers like snowflakes, and they will produce these products within a few days.

By using "small order fast response," Shein has started in the clothing field and gradually expanded into non-standard product areas such as beauty, home, and pet supplies. Shein's revenue has also been growing rapidly, with an estimated $15.7 billion in 2021, and achieving $22.7 billion in revenue in 2022, which has also driven up Shein's valuation.

AI E-commerce Should Not Develop at the Expense of Infringement

Behind Shein's prosperity, some problems brought about by rapid development should not be overlooked. These problems can be summed up in one sentence: "Behind the Shein empire, there is no Coco Chanel or Yves Saint Laurent. Instead, there is a mysterious tech genius."

In this overseas evaluation of Shein, it is believed that Shein lacks a master of design aesthetics but has a genius in algorithms. This has led to many AI-generated products by Shein lacking originality, and even many original designs by designers are suspected of being "plagiarized," leading to many lawsuits against Shein for infringement.

In July of this year, three independent American fashion designers, Krista Perry, Larissa Martinez, and Jay Baron, filed a lawsuit against Shein, accusing it of engaging in "systematic intellectual property crimes" dominated by artificial intelligence. They pointed out that many products on the Shein platform bear a striking resemblance to their original works, indicating infringement.

Original image of "Blooming Flowers" and the product on Shein. Image provided by Business Weekly of Artificial Intelligence

The designers are seeking damages from the court and hope to obtain an injunction to prevent further infringement. Shein issued a statement on July 14, stating that it will take all infringement allegations seriously and take action as soon as possible. "We will actively defend ourselves against this lawsuit and any baseless claims."

It is estimated that Shein will not stop this behavior. An overseas commentator remarked, "If Shein's 'design' 'algorithm' does not generate precise copies that could seriously damage the careers of independent designers, especially since Shein's artificial intelligence is smart enough to plagiarize the most commercially viable works." In short, using AI to "reference" hit products and replicate them for sale is the most effective shortcut.

Not only designers, but also well-known fast-moving consumer brands such as ZARA, H&M, Nike, UGG, Levi Strauss, Kikay Earrings, Stussy, Oakley sunglasses, and Tribe Tropical have all filed infringement lawsuits against Shein. Lawsuit documents show that Shein and its affiliated companies are involved in approximately 100 intellectual property infringement lawsuits in the United States alone.

One notable case is that Quinn Jones, co-founder of the Los Angeles DTC earring brand Kikay, once posted on Instagram accusing Shein of allegedly stealing its product designs, sparking over 1,000 discussions. Many small fashion retailers facing similar issues expressed the same viewpoint.

In fact, AI-generated images also involve dataset training. Whether AI-generated images have copyrights is still not uniformly defined in the industry. Therefore, the current focus of the e-commerce industry's use of AI is mainly on replacing product background images. For example, a certain AI painting platform in Hangzhou has developed an AI e-commerce workflow that can replace product background images in 3 seconds. Additionally, using AI to generate models to showcase clothing, as Alibaba's Lingyang One does, lacks such an advanced platform domestically for designing and generating hit products using AI.

Hidden Concerns in the Future of AI E-commerce

AI e-commerce is demonstrating powerful capabilities. Taking the Shopify platform, which is currently focusing on upgrading AI capabilities, as an example, the latest GMV (Gross Merchandise Volume) for the Q3 quarter is $56.21 billion, a 22% year-on-year increase. From the beginning of 2023 to the present, Shopify's stock price has risen by 72% year-on-year.

Shopify, with its independent station model, leverages AI to enhance merchants' capabilities in decoration, website building, and operational analysis. The Taotian Group, which has many merchants, is also striving to integrate AI into every aspect of its business.

According to a latepost report, the Taotian Group has consolidated more than 20 teams responsible for AI business into 4 teams, each responsible for Alibaba Mama, C-end consumers, B-end merchants, and industry-specific applications.

The Taotian Group has already trained its own large model product "Xingchen." With the support of large models, "Xingchen" serves both search, advertising, and recommendations, as well as contentization for browsing.

On the B-end, the Taotian Group has launched AI tools such as intelligent model image generation, official customer service chatbots, and the Wanxiangtai Unbounded Edition. On the C-end, Taobao has also launched an AI intelligent assistant "Taobao Wenwen," providing consumers with shopping advice, travel planning, and other content.

At the same time, the Alibaba International Business Group has also established an AI Business, with a team of over 100 AI professionals, including approximately one-third dedicated to model training.

Shein is also accelerating the integration of AI capabilities in order to meet the needs of an increasing number of consumers for "small order fast response." According to Vox, Shein is using AI to achieve "real-time retail" in its business model, aiming to complete new designs within three days or even faster in the future. However, it is undeniable that infringement issues arising from AI are still hindering Shein's rapid development.

Currently, infringement issues related to AI-generated images can be avoided. A manufacturer in Shenzhen mentioned that they can use large language models (LLM) and computer vision (CV) algorithms, along with 13 million patents and nearly 3 billion parameters for self-supervised contrastive learning, to provide early warning for product image infringement issues.

However, Shein, which is rushing to go public, may not have had much time to consider these issues. In this regard, it seems that the AI e-commerce model of Shein, which is on the fast track to an IPO, may not necessarily align with the path envisioned by Jack Ma.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。