Original: Ou Yujiao
Source: Home of the Big Model
Image source: Generated by Unbounded AI in less than a year, few companies have relied on the ability to chat and write poems on large models. Nowadays, the combination with the industry has become the new large-scale model narrative…
Recently, the home of the large model noticed that the Jingdong App has launched the beta version of the Jingyan AI assistant.
Source: Jingdong APP
It is reported that Jingyan AI assistant is an intelligent shopping guide service created by JD.com, which can support voice and text input methods, and can provide shopping inspiration according to the content entered by users.
It is worth noting that JD.com emphasizes that any services provided by Jingyan AI Assistant are only used as shopping inspiration, not only do not constitute any substantive suggestions or opinions, but also cannot be the basis for any decision-making purchase behavior.
At the same time, all products and related pictures, videos, and audios contained in the output content of Jingyan AI Assistant do not mean that the platform recommends products and related content at any time, and JD APP is not responsible for the authenticity, completeness, accuracy, timeliness and practicability of the content generated by Jingyan AI Assistant.
At present, users only need to search for the keyword “Jingdong dialect” on the JD App to experience it. However, JD.com has not yet revealed whether the app will be fully launched, and the specific time of launch.
Beijing Tokyo dialect, complete but imperfect shopping guide
After opening the Jingdong App, the home of the large model quickly jumped to the Jingyan AI assistant through the above steps.
According to the content of the page, Jingjing dialect has functional tags such as “professional category consultation”, “personalized gift assistant”, “product comparison assistant” and “shopping experience and knowledge”.
First of all, the home of the large model chose to experience the recommended question under the professional category consultation section: “How to choose whole wheat bread”? About 30 seconds, the Jingyan AI assistant gave the answer. In addition to the answers, it not only pushes links to relevant products so that users can click directly to the purchase page, but also generates a series of related questions about the desired products. It can be said that whether it is the number of answers, the accuracy or the related questions derived, the Jingyan AI assistant has performed very remarkably.
Source: Jingyan AI assistant
However, the large model home noticed that in the current recommendation results, all products are from JD.com’s self-operated stores - JD Supermarket, and the randomness of the recommendation is high (repeating the same question, the product links that appear are not the same), which will limit the comprehensiveness of the answer to a certain extent.
After the professional category consultation, the big model home experienced the “personalized gift assistant” section and chose a controversial topic - namely: what nutritional products to give to grandpa with high blood sugar?
Source: Jingyan AI assistant
However, this time, the Jingyan AI assistant is biased in the key words of the problem. In addition to giving useless answers such as walnut powder gift boxes and nutritious drink gift boxes, the option of sugar-free bread is also given, and sugar-free bread is not a nutritional product. It can be seen that under the premise of concretization and specialization, Jingyan AI assistant will not only provide “supplement” nutrition in the narrow sense, but also provide some food nutrition in the broad sense.
Similarly, in addition to answering user questions, Jingyan AI assistant also extends a series of related new questions based on the questions raised by users. From this point of view, the current Jingyan AI assistant is indeed transitioning from using AI services to AI-based services.
Source: Jingyan AI Assistant
In addition to the common Q&A, the home of the large model challenged the “shopping knowledge and skills” in the form of voice questions. When it comes to the answer to “how to choose a sweeter watermelon”, the Jingyan AI assistant can quickly give the answer in the form of text.
It is true that, on the one hand, judging from the current internal test results, there are still many details to be improved in the Jingyan AI assistant at this stage. But on the other hand, the emergence of Jingyan AI assistant has indeed updated and iterated the original shopping method. This iteration is not only limited to the way consumers search, but also reflected in the sales and service side. “Recommended search” is carried out according to user needs, which improves the accuracy of product purchase and creates a more convenient and pleasant shopping experience for users.
At the same time, with the gradual development of “diversification” and “concreteness” of consumer demand, traditional service methods can no longer fully understand the natural language of consumers, which is very easy to cause user loss and sluggish platform growth. In this context, the birth of Jingyan AI assistant can be said to be at the right time.
Continuing to increase the weight, JD progressively promotes the integration of AI and industry
In fact, the launch of Jingyan AI assistant is not only another new update of JD’s APP AI function, but also represents JD’s continuous consolidation of the detailed performance of industrial AI after leveling the foundation in the era of artificial intelligence.
As the foundation of JD.com, retail plays a vital role in its entire business segment. At the same time, because the marketing and customer service in the e-commerce business have the most urgent demand for large models, and they are the easiest to combine, therefore, as early as 2012, JD began to apply intelligent customer service, focusing on using AI technology to assist customer service to improve human efficiency and optimize customer experience.
In this year’s 618, JD.com is the first time to apply the “Yanxi” large language model technology to official services (including intelligent customer service, AI live broadcast, etc.).
Now, with the advent of Jingyan AI assistant, it not only fills in the gaps in the planning and information modules in the original e-commerce platform, but also completes another integration of JD e-commerce to AI. Therefore, for the current JD APP, backed by a large number of internal testing opinions and demand data base, how to quickly iterate the user search method while realizing the successful integration of AI and industrialization will also become an important issue for it.
In the longer term, as more and more AI genes are injected into JD App, how to give full play to its own full-chain e-commerce advantages to achieve full integration of various business sectors in the context of the deep integration of JD and AIGC is also the key to its successful breakthrough and even writing a new story of the new AI track.
Write at the end
According to the incomplete combing of the large model home, as an emerging technological means, there are currently AI tools such as Taobao Wenwen that have the same path as Jingyan AI assistant and play a role on major e-commerce platforms.
At the same time, the current JD in the financial field and the health field also have AI component stories. Among them, the former has released an AI marketing growth platform based on the application exploration of large models, which has greatly optimized the marketing operation process, reduced the learning cost and operation cost of operators, and improved the production efficiency of the solution by hundreds of times, while the latter has created a large model of Jingyi Qianxun Medical based on the Yanxi large model, which is used in telemedicine scenarios to solve the problem of insufficient supply of high-quality medical resources.
Combined with the JD large model, "for the application practice of the Yanxi model, JD has also formulated a clear “three-step” plan: that is, the large model will be launched in July 2023, internal practice, product integration, tempering and iteration will be carried out in the second half of 2023, and the service industry will be fully opened in the first half of 2024, and the large model capabilities will be opened to external serious business scenarios.
The home of the large model believes that with the continuous stimulation of the demand of the industry, the large model will soon usher in a new round of outbreak, for Jingdong, how to land, how to combine with the specific scene of the industry, will also become a difficult hurdle that it needs to overcome one by one.
