In the context of the booming digital wave and industrial transformation led by new technologies, digitization and intelligence are providing new opportunities for enterprise transformation and upgrading.
On November 9th, Amazon Cloud Technology released three validated scenarios of generative AI in the manufacturing industry, namely the generation of industrial product design concept maps, the generation of marketing image library solutions, and the collaboration of enterprise internal knowledge bases, to help enterprises start from the process and manpower, and improve efficiency.
Empowering enterprises with digitalization has always been a development direction that domestic manufacturing has hoped for, with the core being "how to carry out digital transformation" and "whether to reduce costs and increase efficiency after transformation".
In the digital transformation practice of China's manufacturing industry from "quantity" to "quality", traditional manufacturing faces concentrated challenges such as prolonged technological product innovation, poor activation of overall equipment efficiency (OEE) in factories, low supply chain efficiency and elasticity, and a lack of creating new sources of income.
The vigorous development of technology has brought a window of opportunity for the manufacturing industry to reduce costs and increase efficiency. For enterprises, going to the cloud and using cloud tools have become the main way to solve such problems. In addition, with the arrival of generative AI, manufacturing enterprises are ushering in a new efficiency revolution in three specific scenarios.
New opportunities brought by generative AI
Amazon Cloud Technology emphasizes the three validated scenarios of generative AI in the manufacturing industry, first in the field of industrial product design. Specifically, the traditional process typically involves conceptual design, rendering of the design drawings, and finally the design review process, which requires a large amount of manual participation and involves multiple rework and other challenges, resulting in relatively low work efficiency. The large model has been proven to have significant efficiency improvements in the generation of "concept maps".
Gu Fan, General Manager of the Strategic Business Development Department of Amazon Cloud Technology in Greater China, said, "We emphasize the concept map, which refers to the principle of substitution. The concept map is quickly generated by the big model and then integrated into the entire workflow, so many customers are using and benefiting from it
Haier has teamed up with Amazon Cloud Technology and other partners to create customized generative AI solutions, which have been applied in industrial design scenarios such as new product design, model upgrades, and channel customization. According to data provided by Amazon Cloud Technology, this solution has accelerated the overall conceptual design of Haier Innovation Design Center by 83% and improved the integrated rendering efficiency by about 90%.
The second is in the field of marketing. After the online and offline promotional images are stored, AI is used to quickly generate marketing integration materials that match different channels.
Intelligent knowledge base search is the third major scenario for enterprises to land. Domestic manufacturing enterprises are often personnel intensive, with a large number of employees and a large amount of internal information and data accumulated over the years. Amazon Cloud Technology utilizes generative AI technology to assist customers in building an enterprise level intelligent knowledge base, integrating search engines and large language models, enabling enterprise employees to quickly find accurate and practical content in the knowledge base, effectively improving production and office efficiency.
In addition, the development of generative AI has also brought opportunities for small and medium-sized manufacturing enterprises to overtake on curves.
Gu Fan believes that the similarity between cloud computing and generative AI is that both are infrastructure resources. Faced with the rapid development of generative AI this year, the factors that originally constrained the human and financial resources of small enterprises can be covered by computing resources such as cloud computing APIs, which to some extent has brought small and medium-sized enterprises and large enterprises together.
After solving the technical means, Gu Fan believes that the real difference between small and medium-sized enterprises and large enterprises lies in the business side.
Large model+small model remains the mainstream solution
Since the emergence of generative AI this year, this technology is accelerating its integration into business scenarios in the manufacturing industry, bringing enormous value in product development and design, manufacturing operations, supply chain, marketing and sales, intelligent customer service, and knowledge base.
According to the "2022 Artificial Intelligence Technology Maturity Curve" prediction report released by American consulting firm Gartner, by 2027, 30% of manufacturers will use generative AI to improve the efficiency of product development.
Generative AI has brought greater imagination to the development of various industries, but the current reality is that at the application level, large models are still in a very early stage. Gu Fan is optimistic about the empowerment and efficiency brought by AI to the manufacturing industry, but he is not a fanatical admirer of big models.
A big model is a tool, thinking that a big model is powerful is a kind of technical thinking. A big model is indeed powerful, but it may not be the best choice to solve practical problems. We start from practical applications, and if a small model can already solve business problems and is cost-effective, why rush to build a big model? "Gu Fan said.
Gu Fan believes that enterprises should not focus on the use or non use of large models at the technical level, but more on the application level of scenarios. The large model itself is actually just a tool, and combining different tools into the most suitable scenario to find the optimal solution is the best solution. From an application perspective, using a large model requires comprehensive consideration of whether to accurately address requirements and whether costs are controllable. Gu Fan believes that in the manufacturing industry, "walking on two legs" between large and small models is still the current mainstream solution.