What you learn:
- Adapting the concept of system dynamics to industrial AI initiatives It accelerates adoption and helps organizations avoid mistakes.
- Companies need to collect appropriate data Make sure you are shared with the right people at the right time. Employees need this information to make real-time decisions.
- Don't forget to simplify the frame challenges as a control problemfor example, limit the scope of the project or use a smaller dataset.
Despite extensive access to new technologies, many companies struggle or do not integrate industrial AI into manufacturing. Adapting the concept of system dynamics to these initiatives can help organizations accelerate adoption, avoid mistakes, and improve business outcomes.
Companies need to leverage their data to build feedback loops that lead to more intelligent insights, increased efficiency and ultimately better decision-making. MIT Sloan Senior Lecturer We explore these concepts with new MIT Sloan Executive Education course strategies, survival, and success in the era of industrial AI.
“If you want to be the winner of an industrial AI adoption game, you need to think differently about the system,” Career said in a recent webinar. “Data collection and data filtering technologies, machine learning, and new types of AI can enhance informal feedback loops and are much more successful.
Industrial AI Action Plan
Focus on industrial AI and automation efforts through system dynamics and control lenses, and consider:
Don't aim for a lot of data. Collect appropriate data. Data is central to industrial AI efforts, but excessive low-quality or supplementary data can do more harm than good. Carrier pointed to examples of oil refineries that missed the opportunity to avoid dangerous malfunctions as key indicators are covered in a sea of warning. “They had too much data, so they lost a wealth of information they needed,” he said. “All you need to do is get your data and turn it into information that helps you reduce your risk.”
Strategies, survival, success in the industrial age
Directly on MIT Sloan
Sign up now
Get data in the right hands at the right time. Sharing industrial data via spreadsheets, websites, or cleverly crafted dashboards has so far only done when building useful feedback loops. Real art is pulling these dashboards and signals down to the factory floor, so people at work have the data they need to make real-time decisions.
Cultivate a culture open to data-driven insights. If employees are distrustful of data or are not open to information they don't want to hear, the best key metrics and AI-driven insights are useless. For example, metrics could indicate that maintenance is required or that more training is required. This could call for additional funds or invoke downtime on the line. “A large part of what we benefit from AI is having a culture of accepting data that communicates that we don't want or exclude data that is not currently relevant,” Career said.
Apply control theory to problem solving. It's not an issue of AI technology, but a frame issue as a control issue. Applying quantitative measurements to sequencing processes is an effective way to fully capitalize your system and increase profits and reduce risk. “We recognize that the system is trying to talk to us, and we help by hearing what they're saying and creating that correction feedback loop,” Career said. When applying control principles, it is also important to pay close attention to correct time constraints.
Buy system time. The OODA loop is a decision-making process that helps guide where you invest your time and money to develop a faster response to confusion. Industrial AI data can be used to optimize the four-stage process (observation, orientation, decision, action) especially in the observation and orientation phases where the data may have an impact. Visual recognition and pattern detection with data and AI are especially valuable for the last two acts of the OODA loop that helps detect defects in high-speed production lines and solve problems.
Don't forget that simpler is better. The complexity of AI systems, models, and data can quickly encompass projects and slow down results. Even using small sets of models and data for large-scale implementations of language models, simplifying these systems by limiting the scope of initiatives is a proven way to ensure faster and more meaningful results. “Before you buy these massive technologies that you can go out and do everything, you can focus on the issues at hand,” Career said. “Applying the 80/20 rule is something that really takes to be successful.”
John Career He is an advanced lecturer in System Dynamics at MIT Sloan. He is a field coach who directs senior management on manufacturing and business processes improvements and supports projects. His research focuses on strategic marketing and new business development in the high-tech, specialized chemicals and services segments.
