With the popularization and evolution of AI technology, AIGC's empowerment of the e-commerce industry has been more obvious than ever in the past two years. All platforms, brands, and merchants have actively embraced change and participated in this production tool revolution brought about by AI technology.
In the entire chain of the e-commerce industry, what work content can be replaced by AI?
In what different forms will AI participate in the application process? What are the specific advantages and disadvantages of AIGC's technology in actual interaction?
AI participates in the entire chain of e-commerce
With the technical support of different enterprises and technology companies, AI is now involved in almost the entire chain of the e-commerce industry. The difference lies in the different forms of AI participation in each link.
From early product design to photography, models, copywriting, and art before the product is put on the shelves, whether it is shopping guides or smart customer service, or content seeding and private domain operations...You can imagine that corresponding AI products have been born in most links to help e-commerce people complete their work more efficiently.
In product design, as long as you enter the corresponding prompt (instruction) on Midjourney and Stable Diffusion, you can generate the corresponding product picture. This process is called "Wensheng Picture".
From jewelry and clothing design, home decoration to house interior design, AI drawing function can be applied to almost all related industries. This is also one of the earliest product functions put into use by AIGC. In this sense, the positions that AI may replace areDesign and Drafting.
After having the product, merchants also need to find a venue, arrange for photographers and models to take photos, edit the photos, screen the pictures after receiving them, and then find someone or polish the copy of the product details page themselves... until the product is finally launched.
The cost of this process is high because it requires the division of labor and collaboration of multiple people. In addition to photographers, models, and designers, stylists and various assistants are also involved.
With the advent of AI, some of the work of photographers, models, stylists and copywriters can be replaced by AI.With the help of AI, these tasks can be simplified into "text creates pictures", "pictures create pictures", and "text creates text".
Merchants can use Stable Diffusion (hereafter referred to as SD) to turn the mannequin fitting pictures into realistic model portraits, which can be done by just one person and a computer. Previously, this technology required pre-masking or cutout processing, but after the emergence of AIGC, one-click cutout has also become one of its functions.
Image source: Jirui Technology
In the absence of product pictures, merchants can also directly use the Prompt command in SD or Midjourney to generate corresponding model fitting pictures, achieving "one-step completion".
Apart fromFamousIn addition to Midjourney and SD, domestic companies such as Meitu, Mogujie, and Jirui Technology, which recently received tens of millions of dollars in Series B financing, have all launched products that can "create images from pictures."
ECpro launched by Jirui Technology can generate corresponding product details pages based on product screenshots and complete product entry and other tasks. PhotoMagic can generate real-life model images based on mannequin images, and physical products can complete a cutout. WeShop under Mogujie and WHEE and Meitu Design Studio, products under Meitu, also have similar functions.
Meitu's product Kaipai has expanded its product functions to the field of short videos. It can not only generate marketing copy with one click, but also generate videos recorded by digital human anchors, perform intelligent editing with AI, and package short videos with one click.
AI digital human anchor is one of the most widely used functions of AI technology in the e-commerce field.After the emergence of digital human anchors, the jobs that may be replaced are anchors, assistant anchors, and those related to live broadcast operations.
The AI digital human anchor can automatically explain the products and answer questions based on keyword prompts. This type of digital human anchor will not "collapse" and does not need to rest. After being bought out, it can broadcast non-stop for 24 hours.
Compared with the startup cost of hundreds of thousands of yuan for traditional live broadcast rooms, AI anchors who cost 2,000 yuan a month save merchants a lot of costs for setting up live broadcast rooms. Now many small and medium-sized merchants, brand merchants, and some MCN agencies have begun to use digital anchors.
Image source network
The head of an MCN agency once told the media that now big brands such as Helena Rubinstein are using AI anchors to embrace change, and store broadcast anchors are not very irreplaceable. He predicts that within the next 3-5 years, 30%-50%'s store broadcast anchors will be replaced by AI anchors.
In terms of manufacturers and products, foreign D-ID and HeyGen can generate digital people and short videos from a single photo and provide voice based on text.
Domestic manufacturers, both large and small, are also planning their own AI short videos and digital human products. In addition to the BAT group, NetEase, 360 Group's Xiaoice Company and other manufacturers have successively released their own digital human products or generation platforms. Taking Baidu Intelligent Cloud Xiling platform as an example, the platform can generate automated live broadcast digital humans. The digital humans can also perform built-in movements, and their lip shape and emotions can adapt to text and voice.
Image source: Baidu
In addition to digital live broadcast,Other e-commerce tasks that AI can participate in include smart shopping guides, smart customer service, private domain operations, smart procurement, participation in brand placement analysis, and so on.
At present, platforms such as Taobao, Tmall, JD.com, and Pinduoduo have put intelligent customer service into practical use.
Some cross-border e-commerce merchants used a cross-border e-commerce intelligent customer service product called "Timely Language" to help solve 75% of user problems and increase the conversion rate of manual chat orders from 15% to more than 30%.
AI shopping guides can recommend products based on user needs, gain insight into real user portraits during active searches and interactive Q&A, and improve conversion rates.
AI can also participate in the planning and design of advertising and marketing, produce creative images and short films, help businesses conduct private domain operations, answer user questions in the form of robots among the merchant's user base, guide customers to place orders, and "reduce the burden" for merchants.
From product design to live streaming marketing, from assisting user decision-making to delivery analysis, there are "too many" parts that AI can participate in in the e-commerce field.
AI as an ideal tool
So, as a new "production tool" for e-commerce people, what is the core value of AI besides the expansion of functions?
For practitioners in the e-commerce industry, AI is an ideal tool that can help achieve “cost reduction and efficiency improvement”.
The payment method of AIGC products is often a buyout or subscription system. After paying once or according to different standards (annual or monthly), users can efficiently produce content and use it without restrictions. When dealing with repetitive, mechanical and some creative work, AIGC technology can greatly liberate personal energy.
Especially when this technology is combined with different fields and scenarios, it plays a role in various forms, such as AIGC+pictures, AIGC+text, AIGC+short videos, AIGC+live broadcast... The scalability of AIGC technology is surprising.
Large AI models do not get "tired" and can quickly adjust work results based on feedback. Quick feedback and flexible adjustment are also one of the advantages of AI technology application.
As the cost of use decreases, people's requirements for AI-generated content to achieve conversions also decrease. For example, when the cost of a digital human is one-tenth of the original cost, it only requires AI to achieve one-tenth of the conversion rate of the original real-life anchor.
However, according to feedback from some small and medium-sized businesses, the cost of digital human live streaming is only one twentieth of that of a real person, and the conversion rate of AI can sometimes reach half or even be the same as that of a real person. AI performs well in conversion rate.
During the 618 promotion this year, a certain catering brand merchant used digital human anchors to broadcast simultaneously in three live broadcast rooms. The saved costs were used for traffic investment. Ultimately, the daily GMV of the three live broadcast rooms increased nearly five times compared to that of real-life anchors.
A creative planner from an advertising company once mentioned in an interview that compared with the solutions provided by designers, AI-produced ads can save 30% in costs and 70% in time, making them very suitable for shopping nodes with tight time and urgent tasks.
A Weibo user named "Langzhuhuitou" who works in the e-commerce industry once posted a quote from a model photography studio, in which photography fees and model fees accounted for the majority. In total, the cost of shooting for just 8 hours a day was more than 35,000 yuan.
He said in a Weibo post: "This year we don't need to spend 40,000 or 50,000 yuan a day to hire photographers and models to shoot clothes. (AI technology) is enough for us."
AI products can help many small and medium-sized businesses enter the market smoothly, and can also help large brands and companies reduce costs. At a time when customer acquisition costs and traffic investment costs are constantly rising, it is almost "rarely seen" for e-commerce people to encounter a tool that can "reduce costs and increase efficiency."
AI that is not “intelligent” enough
However, in actual application, AI is not perfect. There are times when AI and large AI models are not flexible or intelligent enough.
The high efficiency of AI-generated content does not mean high availability. AI-generated ideas, pictures or texts sometimes do not meet industry standards and require manual adjustments.
AI sometimes cannot cope with personalized issues. This is sometimes due to technical barriers and sometimes because it does not have enough access to information sources. These problems are essentially due to the information gap and "experience" gap between practitioners and AI.
Large AI models may occasionally exhibit "hallucinations", which refers to the phenomenon that the information source in the content output by AI is fabricated or inaccurate, unreliable, and AI is deceiving itself, which can also cause certain troubles to users.
AI still has the problem of having certain learning and communication thresholds.
Compared with the time, energy and money costs of communicating with humans, ordinary e-commerce practitioners who do not delve into AI technology have a rigid technical threshold for communicating with AI, and this threshold is difficult to cross in a short period of time. Many times they can only seek help from people with professional knowledge. The "cost of use" is reduced, but the "communication cost" may actually increase.
The limitations of AI are not limited only by lack of experience and training, but sometimes also come from external constraints.
There may be copyright disputes over the images, texts, and videos generated by AI. The image and language patterns of AI digital human anchors may make some users feel disgusted and unacceptable. Some platforms have also introduced restrictive clauses related to AI digital humans to limit the functions and use of AI.
A scenario that is closely related to our daily life is the problem that AI intelligent customer service is not “intelligent” enough.
The intelligent customer service of mainstream shopping platforms can now match corresponding service interfaces based on keywords, such as handling returns, exchanges, complaints, and inquiring about service progress. However, when encountering inquiries without corresponding keywords or corresponding entrances, they often appear not to be "intelligent" enough.
Users cannot have a real "conversation" with intelligent customer service and still need human customer service to intervene and resolve the issue.
One oft-cited shortcoming of AI in advertising and marketing is its limited availability.
A designer said in an interview with the media that the content generated by AI-assisted software is still not up to the standard of a finished product, and some copywriting and packaging design do not even comply with advertising laws and design specifications, which poses certain risks. The final effect still needs to be manually reviewed and adjusted before it can be put into use.
In general, AI technology can indeed achieve the effect of "reducing costs and increasing efficiency" expected by merchants, MCNs and other entities during its application, but it also has defects such as being not flexible enough, not smart enough, and having certain technical barriers.
As AI technology is being heavily invested in and applied, some functional deficiencies are expected to be resolved in a short period of time. However, other macro issues such as ethics, technical barriers, and acceptance will have to wait for further development of technology before we can see.
Finally, AI technology empowering the entire chain of e-commerce has become the current situation in the industry. AI technology will become a powerful tool for small and medium-sized businesses to quickly enter the market, and it can also help some cross-border e-commerce merchants get started better.
For most positions, AI can currently play a supporting role rather than a complete replacement, and many jobs still require human flexibility and experience to handle.
As AI technology continues to advance, how can we strengthen our core competitiveness while learning AI as quickly as possible, utilizing AI, and making the most of it?maximumDigitalization may become a long-term issue for e-commerce people in the future.