Digital watermarking technology will usher in national standards. A few days ago, the National Network Security Standardization Technical Committee publicly solicited opinions on the national standard "Information Security Technology Digital Watermarking Technology Implementation Guide" (hereinafter referred to as the standard). The standard defines the technical framework of digital watermarking, and the watermark needs to be embedded in the watermark carrier (such as images, videos, documents, etc.); Clarify the function of digital watermarking technology; Clarify the realization process of digital watermarking; Suggestions on the selection of watermarking algorithm, the use of typical watermarking scenes and the evaluation method of watermarking function are also given.
Recently, Qi Chen, founder of WeShop and chairman of Mushroom Street, mentioned in an exclusive interview with national business daily that AIGC technology has been applied to the commercial auction market, and there are problems in industry standards and norms of AI commercial auction at present. After the establishment of subsequent industry standards and norms, the company will continue to study and follow up.
In 2023, as the founder of WeShop, Qi Chen joined the AI Application Working Group of the Ministry of Industry and Information Technology. After many ups and downs in the Internet technology market, he knows that the high-growth technology industry often goes through the stage of immaturity to chaos, and then to maturity.
In 2023, AIGC technology set off a new wave of innovation and entrepreneurship in the domestic Internet technology industry. Qi Chen felt that AIGC technology was like a hammer, and only by finding a nail that he was good at could he nail down one side of the space. From graphic e-commerce to live e-commerce, Mushroom Street gives people more awareness of the platform of To C. This time, unlike in the past, he aimed at the B-end market and launched WeShop’s AI commercial shooting platform.
How big is the commercial space for AI commercial shooting? National business daily (hereinafter referred to as NBD) reporter chatted with Qi Chen.
AI commercial auction lacks standards and specifications.
NBD: As an e-commerce user, I found on some e-commerce websites in China that some merchants have used AI commercial auction to display and sell goods. If you don’t pay attention, you may not be aware of the use of AI technology, but I saw that the feedback from users in the comment area is AI commercial auction. What do you think of the current application of this technology?
Qi Chen: The point you mentioned actually fed back several questions: First, the current AI commercial auction technology is still in the development period, and there is still room for continuous improvement; Secondly, it also involves issues of industry norms and legislation. WeShop is very cautious in this respect. For example, last year, we were actively applying for the filing of the national AI algorithm. At the beginning of this year, WeShop passed, and we were also the first batch of enterprises to pass.
Recently, we saw the news that digital watermarking technology will usher in the national standard, requiring AI to add digital watermarking to generated content. In fact, before the draft for comment came out, we were also making relevant preparations for AI image watermarking.
We have experienced several rounds of ups and downs in the industry before, and we have some experience, so we think it is good for everyone to make early judgment and preparation.
NBD: Why is the direction of re-starting a business AI?
Qi Chen: In earlier years, Mushroom Street tried the technology of virtual try-on, and AI technology was also applied at that time, but the AI at that time was generally biased towards the recognition function. This round of AIGC focuses more on generative AI. At the end of 2022, when AIGC exploded, we suddenly realized that at this critical point, it seemed to meet some of the needs of virtual fitting that we had hoped to achieve before, so we started the project in early 2023, and about three months later, we went online to shoot WeShop.
NBD: The idea at that time was to make a new iteration on the previous AI recognition technology?
Qi Chen: As far as my entrepreneurial path and main line are concerned, there is still a very clear trunk in general. It is how to show more real goods on the virtual screen, so that the information gap between consumers and businesses can be reduced. The waterfall stream and live e-commerce created by Mushroom Street in the early years are all for this purpose. After this round of AIGC technology broke out, we also tried new projects with this idea.
NBD: Efficiency and quality are actually difficult to balance. Although technology can be pushed forward, can the information gap behind it be eliminated?
Qi Chen: There may be some brands that deliberately create the information gap, but like us, as a retail channel, we hope to reduce the information gap to some extent. From the consumer’s point of view, people may think that consumers naturally want to eliminate the information gap. But in fact, sometimes this is not the case. If there is no information difference between merchants and consumers at all, for example, no packaging at all, consumers may not like it either.
Therefore, our main line is to reduce the information gap as much as possible and constantly use new technologies to reduce the information gap. Usually, when a technology is fully used by the whole industry, its information gap will increase again, so we begin to look for new technologies, which is a process of constant competition and mutual influence.
Find the small incision application scenario
NBD：WeShop was developed by the artificial intelligence team of Mushroom Street. Compared with the previous e-commerce platform of to C, WeShop is a B-side AIGC application, which user groups does it mainly serve?
Qi Chen: WeShop’s business is biased towards the B end, which is a new attempt for us. In addition to the e-commerce platform and retail platform that we tried before, we also hope that as an importer and developer of technology applications, we can apply technology to target customers more widely.
At present, according to our user data, it is mainly divided into several categories: one is domestic e-commerce customers, the other is cross-border e-commerce customers, and the other is brand design customers and advertising customers. Geographically, more than half of our customers are overseas.
An interesting phenomenon is that Brazil is the country with the largest number of overseas customers. Where are their demand points? For example, to sell small commodities and agricultural products, they need to change their backgrounds and models, change their domestic backgrounds to those of Brazil, and change their models of China to those of Latin America.
NBD: In the eyes of many bosses, AI is a sharp weapon to "reduce costs and increase efficiency". From the practical application point of view, what is the willingness of enterprises to use technology to reduce costs and increase efficiency when it comes to specific application scenarios?
Qi Chen: From the perspective of our customers, there is a huge difference between the cognition of the boss and that of the front-line operators. In addition, there are also great cognitive differences among bosses: some people don’t know enough about AI, some people think that AI can solve all problems, and some people think that AI can’t solve anything.
Then, the front-line employees’ cognition of AI is also divided: some people will actively embrace AI and improve work efficiency; There are also some people who will resist and worry that AI will affect their work.
NBD: After the explosive development last year, how do you see the development of AIGC application this year?
Qi Chen: The year 2024 is still crucial. This year should be the stage when AIGC applications are really used by everyone. Therefore, people like us who are engaged in AI application development must do something.
First of all, understand the customer’s application scenarios very well, break down the customer’s entire workflow and work scenarios, and apply different AI functions in different scenarios. Secondly, it is clear in which scenario we can satisfy our customers. At present, ChatGPT is basically completed in the form of dialogue, which seems to be very powerful, but from the work point of view, this interface is actually not professional enough and its controllability is very poor.
WeShop has accumulated hundreds of thousands of customers without advertising and marketing. The reason is that our scene is cut very finely and the granularity is very small. At present, the core of the AI commercial auction launched by WeShop is still staring at the AI application related to the field of e-commerce marketing. From this point of view, 2024 should be a very interesting year. This year, everyone will focus on each segment. In 2023, many enterprises launched large models, some of which cost a lot of money. At first, they were "the hammer of Raytheon", but the actual application scenario to be landed may still be as small as a nail. There may be many times when you don’t need the "Thor’s Hammer" at all. Just find a very precise small incision "small nail" and knock it in with a small hammer.
Therefore, I think in 2024, in addition to "hammers", there will be "pliers" and "wrenches" on the market. The real universal model, just like the industrial machine tool, is to help you make "hammers" and production tools.
NBD: For the current AI commercial auction technology in the industry, some users feel that it has not achieved the expected results. What do you think of the development path of AI commercial auction technology?
Qi Chen: The growth of science and technology is a gradual upward process; Every time the technology advances, more customers buy. Now there are customers who continue to pay and use, which shows that the current level of AI business shooting has met the needs of some customers. However, in the face of the needs of other users, there may be room for improvement in technology, and sometimes we need the blessing of our artificial intelligence team.
I think the technology may have to develop for two years. At that time, everyone is expected to know more about what AI can do, what will be replaced by AI and what cannot be replaced.