The applications using artificial intelligence (AI) have been spreading from consumers to businesses, boosting productivity and strengthening development. With massive accumulation of data, manufacturing has become a blue ocean market for AI adoption. In 2019, AI met the Industrial Internet of Things and the use of AI in industry began.
A global transformation is underway to empower manufacturing with AI. As a world manufacturing hub, Asia has great potential for industrial applications of AI. This report explores the status quo and scenarios of AI adoption by manufacturers to provide insights into the gap between the actual and optimal effects of AI projects and identify industry trends.
During the past century, global technology development was mainly shaped by five prominent trends: electronics, semiconductors, business services, telecommunications, and smart consumer. As the value brought by "Internet + Consumer" reached its peak and then a plateau, a 6th trend—"smart enterprise"—became visible. Smart transformations where companies direct their own solutions using digital technology are the tech trend of the future.
Figure: Global technology trend
Source: Formation & Deloitte Research
Manufacturing is estimated to generate about 1,812 petabytes (PB) of data every year, more than communications, finance, retail and several other industries. As decision-making processes have become increasingly complex due to the surge in digital information over the past two decades, manufacturers have sought to process and utilize information more efficiently, using smart technology to discover data patterns and address problems that could not previously be anticipated.
Figure: Manufacturing tops in volume of data created
Source: GP Bullhound, Deloitte Research
The manufacturing sector has high hopes for AI. According to Deloitte's survey on AI adoption in manufacturing, 93 percent of companies believe AI will be a pivotal technology to drive growth and innovation in the sector. China's performance in AI adoption is outstanding. The market size of AI in the manufacturing sector is expected to exceed USD2 billion by 2025, posting average annual growth of more than 40 percent from 2019. The rise of AI adoption in China's manufacturing sector has been boosted by favorable policies, ample funds, and the potential for AI implementation.
Artificial intelligence can have multiple applications in manufacturing, which can be generally categorized as smart production, products and services, business operations and management, supply chain, and business model decision-making. Smart production is the primary choice for deployment among manufacturingcompanies, followed by products and services. That could change dramatically in the next two years, however. Popular industrial AI applications look set to focus more on products/services andsupply chain management than smart production.
In smart production, artificial intelligence is most used in factory automation, order management, and automated scheduling. Over the next two years, a growing number of AI technologies will be put into use in quality monitoring and defect management, bolstered by advances in computer vision technology.
Although only a limited number of companies have deployed AI in products and services so far, there has been a striking increase in the number of businesses that plan to invest in AI as a priority in the coming two years, with a special focus on applications that shorten design time, customize customer experiences, and enhance marketing efficiency.
There have been a considerable number of AI implementations by Chinese manufacturers. Our survey found 91 percent of these AI projects failed to meet expectations either in terms of their benefits or time invested.
It is quite common to see differences between actual and expected effects in AI implementations due to:
Obstacles from existing experience and organizational structure;
Data collection and quality;
Lack of engineering experience;
Excessively large scale and complexity;
According to our survey, 83 percent of companies think AI has made or will make a practical and visible impact. Among these, 27 percent believe AI projects have already brought value to their companies and 56 percent think these projects will bring value in 2-5 years.
Looking at technology trends, more companies will invest in hybrid technology systems to optimize production, costs, inventory, or quality control, to predict sales and prices, or perform predictive maintenance. Companies are less enthusiastic about investing in technology used for a single purpose, such as visual surveillance, robot localization and expert systems.
Figure: The Technology most popular with the surveyed companies
Source: 2019 Deloitte survey on AI adoption in manufacturing
It is commonly acknowledged that the adoption costs of industrial AI platforms must be reduced for industries to implement AI at scale.
The era of large scale AI implementations in China's manufacturing sector is dawning, and leading companies have begun deployments to gain early mover advantage. Deloitte recommends that companies consider corporate strategy, scenarios, data foundation, teams, partnerships, Proof-of-Concept (POC), and implementation when looking at AI adoption.
Figure: The key milestones in AI projects
Source: Deloitte China Smart Future Research Institute, Deloitte Research
Manufacturing companies need to ensure their AI deployments match their strategies and business goals, be that bringing new revenue, reducing costs, or enhancing operational efficiency. The key is to choose deployments of appropriate complexity to deliver business goals.
Ascertaining where a technology can outperform humans is the proper strategic approach to finding the right AI application scenarios.
AI based on deep learning still relies on big data. A company's data foundations determine whether its AI project will work.
If a company wants to develop AI capabilities, it needs a professional team with AI technology expertise, industry expertise and AI adoption expertise.
After establishing a clearly-defined scenario, complete data foundations, and a professional team, the next step is designing a prototype and running a proof of concept (POC) of the AI process. Iteration and large scale implementation can be carried out if the prototype proves feasible.