The generative AI grand model has not changed
The paradigm of artificial intelligence application in the industrial field
The large model technology represented by ChatGPT and Llama has opened the prelude to general artificial intelligence, and artificial intelligence has become an important driving force for global economic growth, bringing a new space for the intelligence of various industries. According to PWC, AI could contribute up to $15.7 trillion to the global economy by 2030, more than the current output of China and India combined. Of this, $6.6 trillion could come from productivity gains and $9.1 trillion could come from effects on consumption. For the manufacturing industry, artificial intelligence has always been an important part of intelligent manufacturing, Industry 4.0, industrial Internet and other fields, before the rise of ChatGPT, Stable Diffusion, etc., representative artificial intelligence applications such as quality inspection and equipment predictive maintenance have been deeply integrated into the manufacturing industry, and formed a mature application paradigm.
The application paradigm of industrial artificial intelligence has taken shape. First, data science algorithms such as deep learning and reinforcement learning are needed, domain-oriented algorithms such as computer vision, natural language processing, and speech recognition are required, and knowledge engineering such as knowledge graph and expert system is required. For example, models of product appearance inspection are constructed through computer vision, and production planning models are constructed based on reinforcement learning. Construct equipment operation and maintenance service with knowledge graph.
The second is the need for general support technologies to ensure the deployment and reasoning of artificial intelligence applications in the manufacturing industry, such as edge computing, high-performance computing and other technologies to ensure the field reasoning speed, and timing databases, big data platforms and other technologies to ensure the effective management and access of data. Third, industrial knowledge and experience are needed to achieve the adaptation of artificial intelligence applications and industrial scenarios, such as the intervention of expert experience in model training to achieve tuning and optimization, the combination of mechanism models and artificial intelligence models is required in some scenarios to play a role, and the deployment and implementation of models in the production site also needs to be integrated with automated equipment and industrial software.
Figure 1 Industrial artificial Intelligence implementation paradigm
The rise of large models has not caused fundamental changes in the application paradigm of artificial intelligence in the manufacturing industry, but has added specific needs in different aspects. For example, at the algorithm level, basic models based on Transformer, U-Net and other architectures have become the basis for generative artificial intelligence to enter the manufacturing field. In the field of general support technology, vector database and MaaS have also become important digital infrastructure; In the field of industrial knowledge and experience, different from the past demand for structured data such as time series, the requirements of generative AI for high-quality text, pictures, documents and other data are constantly improving. Although the large model is still applied under the original paradigm, the large model technology will continue to expand the application space of artificial intelligence in the industrial field, according to Accenture estimates, Al can increase the added value of the manufacturing industry by nearly 4 trillion US dollars in 2035, according to Marketresearch forecast, by 2032, The global generative AI manufacturing market will reach $6.398 billion.