Few industries are better poised to benefit from applied AI capabilities than manufacturing. With the proliferation of sensors and networks across the operating environment, manufacturers are swimming in data. That data is more valuable than ever amid the challenges manufacturers now face from the COVID-19 pandemic.
Manufacturers are constantly working to identify patterns and solve problems, knowing that even the smallest improvements have big implications. They have always been pioneers in making smarter use of automation, so it seems logical that the automated learning that characterizes AI would find a natural affinity with manufacturing. Yet even with that clear synergy, manufacturers have often faced challenges to AI adoption.
With the difficulties we're confronting today, there has never been a more important time to take full advantage of AI. The answers that AI holds for manufacturing are well-suited to helping them adapt to the pressures and topsy-turvy conditions the pandemic created.
In a recent global survey about AI, responses from more than 300 executives across industries pointed to growing evidence that we are on the verge of a momentum shift. Conducted by Forbes Insights and sponsored by SAS, Intel and Accenture Applied Intelligence, the survey reveals that AI deployment has gone beyond discrete use cases or experiments and into enterprise-wide adoption. Even with gaps in capabilities and strategy revealed through the responses, respondents indicate widespread AI adoption just around the corner – in manufacturing and every other industry.
Our report, AI Momentum, Maturity and Models for Success, reveals
that leaders and early adopters in AI are making important advances and are identifying and expanding on what works as they use AI in more ways and more parts of their organizations.
“The truth is that large global manufacturers still rely heavily on older, disconnected machinery”, says Marcia Walker, principal industry consultant, SAS global manufacturing industry practice. “So, while many manufacturers are using AI, in this survey we see that they’re using it in unexpected ways – in customer-facing operations, for example, or in the areas of warranty claims and recalls. There’s still so much that manufacturers will be able to do with AI as their businesses evolve and they continue to invest in modernizing other parts of their businesses. The good news is that we’re finally seeing AI move into aspects of manufacturing that have remained analog for years, such as on the factory floor. In production, for example, my own experience shows that image recognition is being more widely adopted.”
So, what do manufacturing executives report about their experiences with AI? Where are they focusing today? What’s working? What are their plans for the future?
Remarkable results are possible with AI
Twenty-six percent of manufacturing respondents report that AI-based technology has been deployed, and 50% say it’s under development. These figures are roughly in line with other industries such as consumer packaged goods and retail. This suggests that the manufacturing industry has embraced AI. But, looking more closely, squaring these positive responses with actual experiences in the industry proves difficult.
According to Walker, new manufacturing customers often describe themselves as using AI. But an examination of which AI technologies they’re using (and how), usually shows they’re using AI as a catch-all term for everything from simple dashboards to analytics, basic statistics and rudimentary software-aided automation.
Think of AI as the science of training systems to emulate human tasks through learning and automation. Many manufacturers simply haven’t progressed to that level yet. But when they do, the results can be remarkable.
Given how manufacturers have embraced IoT, the opportunities from applying AI to IoT data seem exponential. “Machine learning and artificial intelligence are areas we’re emphasizing heavily now,” according to Conal Deedy, director of connected vehicle services for Volvo Trucks North America. “We’re uncovering hidden insights in our data and merging that with the truck knowledge from our engineering group. Together, we are in a much better situation to understand exactly what the data is telling us and integrating it into the remote diagnostics service. We are already seeing the benefits and the future is extremely exciting. We can now process millions of records in real time, expanding Volvo’s remote diagnostics capabilities, which on average helps reduce diagnostic time by seventy percent and repair time twenty-five percent.”
Culture change is a critical enabler for AI adoption
When asked to identify the central challenges to successfully implementing and applying AI in their organizations, manufacturers rightly pointed to a range of issues, from ensuring AI-based outputs are objective and neutral (24%) to a lack of development/deployment expertise (26%) and more. And, two related responses stood out – organizational culture (22%) and resistance from employees due to concerns about job security (16%).
Over the years, one of the biggest recurring themes is related to AI is convincing manufacturing employees that the AI systems are as reliable as their gut instincts. A big part of the opportunity for manufacturers in relation to AI will involve creating the right conditions for the cultural changes that will help AI adoption take root.
As is often the case, success breeds more success. One way to foster a culture change is to start with a win.
For example, a European-based manufacturer was pursuing an aggressive plan for new AI initiatives. At first it seemed to make sense to pick an initial AI project that had the greatest potential benefit. But instead, a lower ROI initiative was suggested for one important reason – it was the project most likely to win over skeptical engineers and provide proof that AI works. That early-win AI project convinced the engineers that AI could deliver reliable, trustworthy results, paving the way to implement higher-ROI projects based on the confidence that AI can work in their organization.
Bridge the IT and operations divide to win on AI
It should come as no surprise to any leader working in manufacturing that alignment between business objectives and IT presents a huge challenge to AI implementation. Twenty-two percent of survey respondents pointed to this as one of the most pressing problems they face.
Over the years, operational technology (OT) has become more specialized and sophisticated. And despite important advances toward standardization, these teams still have difficulty communicating with one another, much less agreeing about IT infrastructures. After all, the engineers who designed OT and IT capabilities tend to come from different engineering fields whose systems were designed to solve different problems – software engineering (IT) or mechanical engineering (OT).
Why is this OT-IT collaboration important for AI projects? Because in a manufacturing environment, AI should be able to operate at the intersection of OT and IT. That will not only require enhanced systems learning to communicate with one another, but also deep collaboration between different types of IT and OT leaders as well.
Today, many manufacturers simply aren’t there yet. But there are hopeful signs pointing the way to greater IT-OT collaboration in the service of AI in manufacturing. For example, the continuing advance of cloud capabilities has already forced manufacturers to standardize in ways that open the door to similar approaches for AI.
What’s next with AI in manufacturing?
It seems clear that AI will continue to expand among manufacturers, as leaders become more adept at deploying these capabilities, and as the capabilities themselves become even more powerful. While what’s next will vary from facility to facility based on their operating environment, it’s a safe bet that the factors that set successful AI adopters apart will figure prominently among manufacturers. These factors include:
- Process maturity. Manufacturing leaders should be regularly reviewing AI output and ensuring they have processes in place for confirming or overriding questionable results. Do they have plans to significantly improve business processes using AI? These are indications of process maturity in AI – and they’re all areas in which AI leaders in manufacturing are already setting themselves apart from the pack.
- Connecting analytics to AI. Manufacturing analytics drives the learning and the automation aspects of AI. This connection may not be apparent for manufacturers that have yet to deploy AI. Successful AI users show a level of analytical maturity, and the research suggests that is a key enabler of success.
- Trust in AI. When it comes to AI, success breeds confidence. Successful organizations are more than twice as likely to trust their ability to ethically use AI technologies in the future. Manufacturers are no exception.
- Healthy levels of AI oversight. Companies that have been more successful with AI tend to have more rigorous oversight processes in place. For example, 74 percent of successful companies report that they review their AI outputs at least weekly. While manufacturers appear to have made good progress in this area, keeping up with AI advances and deployments in your organization will require focused attention and effort.
It’s clear that AI deployment is accelerating and only just getting started. If AI were a five-stage rocket, we may be firing the third stage now. And as AI continues its ascent, many of the issues examined in the survey will grow in importance, entering more boardroom-level conversations, landing on more implementation-level meeting agendas and appearing more frequently in media accounts.
- Three steps for conquering the last mile of analyticsPutting your analytical models into production can be the most difficult part of the analytics journey. It’s no surprise that this last mile of analytics – bringing models into deployment – is the hardest part of digital transformation initiatives for organizations to master, yet it’s the most crucial.
- Shut the front door on insurance application fraud!Fraudsters love the ease of plying their trade over digital channels. Smart insurance companies are using data from those channels (device fingerprint, IP address, geolocation, etc.) coupled with analytics and machine learning to detect insurance application fraud perpetrated by agents, customers and fraud rings.
- A guide to machine learning algorithms and their applicationsDo you know the difference between supervised and unsupervised learning? How about the difference between decision trees and forests? Or when to use a support vector algorithm? Get all the answers here.