Marco Santos is co-CEO of GFTa global digital transformation company. I write about technology, payments, leadership and sustainability.
“Is it too late for manufacturers to implement AI?”
“How do manufacturers train their employees to use AI?”
I recently received these questions from a reporter covering the manufacturing sector. My first thought was that many manufacturers are falling short of training their employees. They’re still trying to figure out how to train their AI systems – if they’ve even introduced AI at all. My second thought was that people’s impressions of how AI is used in manufacturing (and other industries) are probably at odds with the actual situation in the industry.
Although manufacturers know that AI presents a significant opportunity for automation and efficiency, many are still in the early stages. first steps in implementing AI. Others, like Fordhave introduced compelling AI use cases that are already improving quality control and decreasing recalls.
The reason AI adoption is so widespread among manufacturers is not because businesses don’t see the potential. Indeed, successful deployment of AI requires fundamental changes to the way they store and manage data across machines, services and sites, often around the world.
It also requires infrastructure built from the ground up to support new use cases – some they can predict and others they haven’t even dreamed of. It all starts with a holistic AI strategy. Training employees on new systems and processes comes much later.
Based on my work with some of the world’s largest manufacturers, here’s an overview of what goes into a thoughtful deployment of AI.
Create the data infrastructure that will support AI.
One of the biggest benefits of AI for manufacturers is its ability to synthesize large amounts of organizational data (we’re talking decades) in a fraction of the time it would take humans to do so. However, achieving this depends on the quantity and quality of data that AI systems can access.
For many traditional manufacturers, this is a daunting prerequisite, given that data consolidation means collecting data from multiple disparate locations within the organization. For example, each machine in each factory can run with its own software from which manufacturers must extract data. They must then link it to data from other machines and data sources in factories and corporate IT systems elsewhere in the company.
Only then can they create and implement a new data infrastructure, either in the cloud or in another digital format, capable of keeping all this data in one place and making it accessible to any tools or systems anywhere in the organization.
Not surprisingly, this basic approach can take years, and many manufacturers are still just getting started. German medical packaging manufacturer Gerresheimer, for example, announcement that it plans to connect its application and production servers to create a centralized data repository.
The good news is that once this first step is taken, manufacturers rarely have to start again. I know from working with manufacturers that half the time they spent on data analysis projects was spent collecting data, while the other half was spent installing and configuring of AI and machine learning applications. Those who have since introduced new data infrastructure are now devoting all their time to creating and deploying new AI and ML use cases.
Create AI use cases that put data to work in the manufacturing process.
AI use cases will be different for every organization, but some will likely benefit manufacturers across industries.
For example, in collaboration with one of the largest automobile manufacturers in the United States, we use visual AI to automate process monitoring in the factory. The company is now able to analyze the precision of parts produced on a stamping line that presses 900 parts per hour. This creates a wealth of data, including images of every part produced. This data is then used to further optimize the process, improving efficiency and quality as AI becomes smarter.
Visual AI can also be leveraged to inspect materials and other products on the machine as they are produced. And when combined with robotic technology, it can go even further, removing any defective parts it identifies from the assembly line in real time. This significantly reduces the time factory workers would otherwise spend on these processes.
By leveraging their extensive historical data, manufacturers can also deploy AI for predictive maintenance. By comparing current machine performance to known patterns of healthy and unhealthy machines, AI can identify performance anomalies and predict equipment issues before they occur. This can ultimately lead to less machine downtime and therefore higher production.
Employee training is essential, on time.
To answer questions I’ve received about how manufacturers are training their employees to use AI; most of them are not yet. The majority are still focused on achieving the previous two areas of their AI journey.
Once they are ready to deploy their AI use cases and focus on employee training, they should focus on two critical areas:
1. Teach employees how to maintain data quality.
2. How to analyze and interpret it effectively when using AI tools.
These are two skills that even the most tech-savvy employees simply won’t have because they haven’t yet worked with manufacturers’ specific AI solutions.
Employees who will directly interface with the AI itself must additionally have a strong grasp of rapid engineering. In other words, they need to know how to write generative AI inputs that will produce the desired result. This skill requires equal parts training and practice.
These new skills will create an environment where employees are ready and able to begin overseeing new AI systems and tools as soon as manufacturers deploy them.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Am I eligible?