Bridging the Gap Series: The strengths and weaknesses of AI
AI is having a major impact on almost every vertical and every aspect of business, including industrial automation. But we need to walk before we can run. What are the strengths and weaknesses of AI, and where can it provide benefits over traditional industrial automation without AI?
Artificial intelligence (AI) insights
- AI excels in complex decision-making by deciphering vast data sets, correlating variables, and predicting trends, enhancing adaptability and problem-solving.
- Integrating AI into industrial automation faces challenges like talent scarcity, outdated technology, and data quality issues, requiring upskilling and modernization.
Regardless of the industry you work in — engineering, building automation, automotive, filmmaking — artificial intelligence (AI) is likely already impacting your day-to-day job. That is absolutely the case in industrial automation. CFE Media and Technology recently released new research, covered in the Bridging the Gap Video/Podcast Series, that took a closer look at how AI is impacting several aspects of industrial automation. In the third episode, we discussed the strengths and weaknesses of AI and how it is practically being used in industrial automation. What does AI do well, what doesn’t it do well and where could it be better used?
AI and complex decision making
Automation is rule based, which means it’s fairly easy to understand how it works. The biggest benefits are that it’s generally explainable, reliable and predictable. If A, then B. AI, on the other hand is not necessarily as reliable. It is designed to make its own decisions so it’s harder to predict what answer it’s going to come up with.
The thing that AI is good at, according to Bridging the Gap guest Jeff Winter, an Industry 4.0 and digital transformation thought leader and influencer, is adaptability.
“AI can make a lot of sense of a lot of things that traditional automation and traditional logic will struggle with,” Winter said. “If your data isn’t perfect in traditional logic, you’ll get errors, versus AI is even smart enough to eliminate those data sets or to potentially make sense of them.”
It can also help with complex decision making, especially when you have several variables to deal with.
“Imagine having, like in the case of some companies for demand forecasting, 3 million different data points that you’re pulling in together from external sources, and you need to identify what’s correlated with what, what are the most important things to pay attention to,” Winter said. “AI is fantastic at that. When it comes to classification or regression or prediction, those are three things that you really can’t rival it and those are the things that it is great at.”
Challenges of AI integration
When it comes to implementing AI into traditional industrial automation processes, many people worry about the technological challenges, but the bigger issue is the talent shortage, Winter said. Companies need to design programs to train and upskill their employees to understand AI.
“If you look at the speed at which AI has taken off, the world hasn’t been able to adapt to that,” Winter said. “Think of how long it takes your education to adapt from high school education to college education to make sure that the industry is prepared for a lot of these technologies. We’re just not ready yet. If you were to go search for the top jobs in demand right now, the most unfilled are related to data science, computer science and AI specifically.”
The availability and quality of data can also be an issue in AI integration, especially in the manufacturing environment. If you look at IT as a function, hardware gets replaced every few years. With manufacturing, that’s not the case. Some machines can be up to 100 years old, and companies may be running programmable logic controllers (PLCs) that are 20 or 30 years old.
“The manufacturing landscape has a significantly more wide range of technologies, protocols and different platforms that you have to integrate with just to get the data at all, let alone having the data contextualized and normalized in a way that makes sense for you,” Winter said.
Bridging the Gap series
Other episodes in the Bridging the Gap series will focus on the challenges of AI integration, trust in AI and the business impact of AI. As companies progress on their digital transformation journeys, AI will become more intertwined with every aspect of business.
Because AI is a software, it does need to integrate with something. As a result, companies that are more automated will be better suited to take advantage of the benefits of AI.
“You can’t improve your processes if your processes aren’t automated,” Winter said. “If you’re looking at how you’re evaluating the overall success, they shouldn’t be mutually exclusive. They should be looked at as how can we take our automation processes today and make them better through the use of AI because we have all the controls in place?”