It has been observed that Industrial AI adoption is slower than the adoption of AI in other kinds of business, like healthcare or telecom; manufacturing industries are slower by nature when making decisions and adopting new technologies; technology life cycles are extended by 10 to 20 years or more; and they are less tolerant to error. The differences are due to high engineering knowledge in industries, a lot of knowledge that has been applied and learned over centuries in this specific domain and all of this knowledge is in the heads of people who think in a specific way and these people are ultimately responsible for industrial operations.
There are two distinct mindsets and ways of thinking: the Data Scientist, who thinks in terms of tables, graphs, decision trees, and machine learning results, and the Process Engineer, who thinks in terms of process diagrams. This brings us to today’s contentious battle between Information Technologies (IT) and Operational Technologies (OT), but if we look closely, we can see that these two branches of technology are fully cooperating. OT is adopting IT at their own pace, but in the end, there is an adoption of IT in instances of Distributed Control Systems, Manufacturing Execution Systems, and all Enterprise Resource Systems. Adoption is also being driven by the Industrial Internet of Things, advanced process control systems with real-time optimization, and many other technologies.
In the near future there will be some enabling technologies that will be supporting Industrial AI, and they are Deep learning, Knowledge Representation and Cloud/Edge computing.
