At the beginning of 2023, the emergence of ChatGPT has set off a new round of artificial intelligence application boom at home and abroad.
In just 5 days after its launch, ChatGPT gained 1 million users; Less than two months after its launch, the daily activity of ChatGPT users exceeded 10 million; Relevant data shows that by the end of January 2023, its monthly user activity has exceeded 100 million, making it the fastest growing consumer-grade application app in history.
According to relevant statistics, the current core industry scale of AI in China exceeds 400 billion yuan, and the number of enterprises is close to 4,000, driving the scale of related industries to exceed trillions of yuan, and will further increase rapidly in the future.
However, both garment companies and manufacturing companies have not suffered losses in the supply chain. For example, during the epidemic, the risk of supply interruption of raw materials and supporting products, the sharp increase in inventory pressure, and the inability of the original supplier channels to take into account the requirements of supply chain operation efficiency have made it difficult for enterprises to move forward.
Although ChatGPT introduces a model of "artificial annotated data + reinforcement learning", it can better understand human intent and assist humans in completing a series of tasks. But in the field of supply chain, how can ChatGPT technology be empowered to accelerate the implementation of AI business?
A disruptive shift from "efficiency tools" to "production tools"
Before ChatGPT, the role of most AI intelligent robots was to help enterprises improve efficiency and reduce labor costs as much as possible, especially at the supply chain end, in order to better reduce costs and increase efficiency, all walks of life are moving towards digital transformation. Through the integration and innovation of emerging technologies such as big data, artificial intelligence, and 5G, and all aspects of the supply chain, we aim to create new value and growth points.
Taking cross-border e-commerce as an example, not knowing English, not being familiar with overseas culture, and not understanding the needs of overseas consumers has always been a problem for Chinese enterprises and European, American and even global consumers. However, ChatGPT directly reduces the threshold and cost of cross-border content through independent production of content, such as writing daily greeting emails to users of independent sites, and doing non-complaint dialogue and communication with consumers, which can be completed by ChatGPT.
Not only that, as a "production tool", ChatGPT replaces more simple and repetitive tasks, organizational structures that used to require a lot of manpower may be disrupted by elite teams, and product and supply chain capabilities become more important. For example, the collaboration network of the entire cross-border e-commerce industry will change, and the entire supply chain system will face slimming and improving overall efficiency.
For brand owners, understanding consumer needs is one of the essential links, but most domestic brands do not understand overseas markets and consumers, even if they rely on foreign e-commerce platforms such as Amazon to go overseas, it is difficult to achieve product focus. And ChatGPT provides a low-cost, access to overseas consumers. No matter what questions are raised, in theory, ChatGPT can sort out the corresponding answers from the information in the current 175 billion database, making the supply and demand docking of global e-commerce more efficient.
For clothing companies, problems such as serious inventory backlog and difficult to improve net profit have always been difficult to solve. Some enterprises have established a flexible supply chain to build consumer labels, commodity labels, and channel labeling systems, and then continuously optimize indicators such as replenishment availability and inventory-to-sales ratio.
With the help of ChatGPT technology, enterprises can more quickly forecast orders through massive data, carry out targeted and planned production, reduce the backlog of enterprise inventory, and design and content planning according to business needs to help find weak areas that need improvement.
For example, on the consumer side, clothing brands can use ChatGPT to automate the design of consumers' body shapes, preferences and styles, and then complete production, fundamentally changing the manufacturing methods of the traditional clothing industry. At the same time, it can also carry out fast and effective production and sales of products in terms of automated production lines and automated inventory management.
From the improvement of the B-end process to the C-end consumer experience, ChatGPT has gradually completed the change from "efficiency tool" to "production tool" in the entire supply chain. TBanicDate estimates that by 2025, the proportion of data generated by artificial intelligence will reach 10%, and ChatGPT products are expected to gradually realize human-machine collaboration in the future, making the application of new scenarios faster, accelerating the intelligent upgrade of various industries, and achieving growth in marginal utility.
Reduce communication costs and facilitate task coordination
Whether it is based on artificial intelligence, big data or emerging digital technologies such as ChatGPT, the underlying logic of the supply chain has not changed, and it is still through linking people, things and information in the supply chain to build an ecosystem of efficient collaboration, rapid response, dynamic intelligence in multiple links such as product design, procurement, production, sales and service.
But generative AI like ChatGPT can proactively generate content and build an omniscient brand image. At this stage, artificial intelligence can only screen information and cannot make decisions, for example, in the sales link, all walks of life can only push thousands of people, but can not accurately grasp consumer preferences; ChatGPT can quickly understand consumer personality, consumption habits and other information through interaction, and carry out product delivery and service.
If ChatGPT is used to train logistics chatbots, ChatGPT can process orders in real time, update warehouse locations, and further improve warehouse management efficiency, starting from receiving orders on the computer and tracking the completion of the entire order.
In terms of improving the level of logistics risk assessment and management, ChatGPT can effectively avoid potential conflicts or known bad weather and select the best transportation route; Help select suppliers based on financial strength, historical performance, infrastructure and other conditions to select the best suppliers. Especially at the cross-border logistics level, involving many issues such as tax declaration and customs clearance, the choice of suppliers is more prominent.
Through the empowerment of emerging digital technologies such as ChatGPT, data throughout the supply chain can be seamlessly accessed, further breaking down data silos between partners, bridging the gap between brands and their external networks, and ultimately enabling multi-enterprise collaboration. Let the entire supply chain system become a richer and broader ecology and promote the future development of enterprises.
For example, NVIDIA's chip research and development, is through big data and other models, self-developed CUDA (General Refrigerator Computing Platform) suitable for the development of deep learning, and replaced the CPU with GPU to become the first choice in the AI training market, and finally NVIDIA became the absolute hegemon in the field of AI chips, and continued to cultivate in the downstream industry chain, in addition to the game industry, the underlying resource support of intelligent driving has also made a huge breakthrough.
Some institutions predict that in the next five years, 10%~30% of the pictures will be generated by AI, which is expected to create a market space of more than 60 billion yuan, and the market size of AIGC will exceed one trillion yuan. However, at present, the industrialization of AIGC is limited, and a large number of business scenarios have not found a monetization model, and it will still be in the commercial exploration mode for a long time in the future.
Future opportunities and risks
Although the explosion of ChatGPT is, to some extent, a milestone event after years of technological precipitation of AI, the reason why ChatGPT can achieve such a powerful interaction is inseparable from the huge computing power support behind it.
Relevant data statistics, the total computing power consumption of ChatGPT is about 3640PF-day, according to a domestic landing of a data center as a reference, to build equipment to support ChatGPT operation, 7~8 such data centers are needed, and the investment in infrastructure alone is nearly 10 billion yuan.
However, computing power is the foundation of AI, in order to continuously improve the threshold of AI, on the one hand, it requires huge investment to use the computing power center, on the other hand, it is necessary to make a large model, and the demand for high reliability, high performance and high security computing power is more prominent. In addition, data is also a key factor in determining the development of AI, and the difference in the quantity and quality of data has a greater impact. Especially in the supply chain field of mainland China, the fragmentation of data sources will affect the generalization ability of AI models and the generalization effect of different scenarios.
In the field of supply chain, the upstream, middle and downstream enterprises involved are different in nature, and their respective data sources are relatively fragmented, according to the company's main business, the scope of data collected is limited, and the limited data collection methods affect the quality of data that can be used for model training at the multi-dimensional level. For example, Tencent collects social data through WeChat, consumer behavior and decision-making data through fresh retail, and so on.
However, in the neutral nature and open style of OpenAI, it is easier to connect the upstream and downstream of the industry chain, and compared with the existing AI robots in China, ChatGPT has more advantages in open ecological construction, far better than any single platform company's data source. Even so, ChatGPT can't replace humans in making decisions.
From a technical point of view, although ChatGPT realizes the versatility and high interaction of artificial intelligence, it still does not have the ability to think about extension, and needs to be trained and recognized with the help of RLHF of human feedback, and cannot make professional answers in professional fields such as medical treatment and physics.
From a security perspective, ChatGPT is currently in the "usable" phase and has not reached the "trusted" stage. On the one hand, because the content produced is not original, but through the association statistical relationship between words, and then synthesize language answers, content security and accuracy are difficult to guarantee, on the other hand, due to the data security and copyright issues of related content, it has not been effectively supervised, and is still in a gray area. For example, the use of ChatGPT for the writing of doctoral dissertations, etc., although some university alliances in the United States, Europe and other countries have explicitly banned the use of ChatGPT in academia, but have not made relevant penalties.
In addition, there are many risks and hidden dangers in the collection of data between countries, especially in China, where there is a risk of information security issues such as information leakage.
That being said, some agencies say that ChatGPT's influence has fulfilled an important sign of revolutionary technology and could lead to a large number of applications redoing existing lifestyle changes. For example, AI access equipment, release a lot of productivity, or change the habit of using traditional search engines, innovate the content industry, and so on.
In the long run, ChatGPT may be able to lead the technological revolution and reconstruct the global supply chain pattern.