Nearly 20 years ago, Director Wu of the equipment workshop at Baosteel's continuous casting branch had an idea: extract the data from the continuous casting equipment and transmit it via the network. This way, the equipment status could be monitored from the office. Looking back, this is a typical example of the Industrial Internet concept. However, I wasn't enthusiastic about this idea and didn't participate. The reason is simple: while this approach might make things more convenient for inspection workers, the workload would be quite substantial. I felt the return on investment wasn't worthwhile. According to the company's performance evaluation system, doing this kind of work would essentially be a waste of time and effort.
In fact, the key to enterprises' confusion about the Industrial Internet lies in the word "value": What exactly can the Industrial Internet do?
More than 20 years of innovation work has made me accustomed to considering value from the perspective of "opportunity": What things were impossible or unnecessary to do in the past, but are possible and necessary today? Opportunities are often hidden there. Therefore, my thinking often begins with reviewing history.
It is well known that large enterprises typically have quite sophisticated computer systems. However, these systems are often hierarchical: lower-level computing systems are generally called automation systems (such as Business Architecture (BA) and Process Control Systems (PCS), while higher-level systems are called information systems (such as Enterprise Resource Planning (ERP)). Lower-level computers handle mostly real-time problems, with a very high degree of automated control; while higher-level computers handle problems with relatively lower real-time requirements, with a very high degree of manual management. However, automated control systems have a very small impact, such as a single valve or piece of equipment; while information systems have a much larger scope, potentially managing an entire workshop or factory.
In principle, there is information communication between computer systems at different levels. Higher-level computers are responsible for generating goals and requirements, which are then distributed and broken down to lower-level computers for execution. The execution status of lower-level computers is also uploaded to higher-level computers so that managers can monitor production. However, this uploading is selective; generally, only production results or specific problems are transmitted. Uploaded data rarely triggers automatic feedback and adjustments from higher-level computers.
So, the question arises: where do the opportunities lie for the Industrial Internet?
In my view, this opportunity lies in the contradiction mentioned earlier: the underlying system has strong real-time performance but a small scope of management, while the upper-level system has poor real-time performance but a large scope of management.
Theoretically, automation is ideal. We know that one function of automatic control systems is to replace some human work. But their significance goes beyond that. For example, the response speed of automatic control systems can easily reach the millisecond level, while the human response speed is much slower. Therefore, the control effect of automated systems is often better than that of humans, and their value in improving quality and reducing costs is more important than reducing labor costs.
Given this, why can't decision-making in information systems be automated? A key reason is that information systems often face problems far more complex than automated control systems. For example, an automated control system functions only when the equipment is functioning correctly, but when a piece of equipment on the production line malfunctions, the manager must decide what to do. Some problems require human expertise to handle flexibly. Management systems, in essence, are systems that help humans handle problems flexibly.
People have strengths and weaknesses.
Equipment information can be collected in milliseconds, and a factory may have tens of thousands of devices. This equipment information includes not only equipment status but also human operation and management data. If there is enough information, various issues related to quality, efficiency, and cost can be identified from the equipment data. However, humans are not infallible. If all this data were sent to humans, their real-time processing capabilities would be insufficient to manage it all.
What do we do when our brains aren't big enough?
GE's Industrial Internet white paper identifies three key elements: intelligent machines, advanced analytics (algorithms), and (online) workers. "Advanced algorithms" essentially allow computers to act as human secretaries; these secretaries manage and analyze data, relaying problems to humans when they cannot be solved. This allows humans to better manage factories with the help of this "external brain." Collaboration between humans and algorithms is crucial for handling issues with significantly different timeframes. In fact, the subtitle of GE's Industrial Internet white paper is "Pushing the Boundaries of Minds and Machines," which can be understood as "reconstructing the interface between humans and machines," embodying the collaboration between them.
In particular, machines, as "secretaries," possess many capabilities that the human brain cannot match: an exceptionally strong memory. Collecting historical data from thousands of devices would constitute a vast intellectual asset, something the human brain simply cannot process. Five years ago, when I invited Mr. Salvo to Baosteel to speak about the Industrial Internet, he mentioned that it could help people operate in parallel during product design, making product design and R&D processes more intelligent. Only recently have I truly understood this concept.
When humans and "advanced algorithms" collaborate, it's like a pedestrian boarding an airplane. This significantly enhances a company's management capabilities. I've repeatedly emphasized that "management defines the boundaries of technology." With improved management capabilities, a company's competitiveness naturally increases: costs can be significantly reduced, quality and efficiency can be significantly improved, redundancy disappears, and the corporate culture becomes healthier. The value is immeasurable. In this way, the perplexing question of the "value problem" of the Industrial Internet is resolved.
Let's go back to Director Wu's idea 20 years ago. At that time, not only was I uninterested, but the company wasn't either. If we go back 20 years, was there a way to get the company interested? Yes! The way was to help the company clearly explain its strategic thinking. The value of doing this shouldn't be viewed solely from one perspective, but rather as becoming the world's most competitive steel company and cultivating Baosteel's future competitiveness in the IT field. In this light, doing this might become a task that everyone values highly. Of course, the chosen entry point doesn't necessarily have to be continuous casting.
I often say that innovation often originates from personalized needs. Great innovations are often "the beginning of a hurricane." Now it seems that without looking at problems from a strategic perspective, the wind cannot blow.
In short, the functions of the Industrial Internet differ from those of traditional enterprise automation and information systems. This difference is where the opportunity for the Industrial Internet lies. In the short term, Industrial Internet systems should perhaps operate in parallel with existing enterprise systems, collaborating and dividing tasks. This helps reduce the risks associated with system upgrades and replacements. When conditions are ripe, integrating the functions of various systems into a unified system is also a realistic possibility.
