In April, the Detroit Society for Coatings Technology held its annual FOCUS conference in Livonia, Michigan. The 2024 theme, Emerging Challenges, Sustainable Solutions, included topics such as automation/digital, sustainability, anti-corrosion, specialty coatings, and color. OEMs including Ford, GM, Stellantis, and Boeing joined coatings manufacturers and suppliers for a day of technical presentations, specialized workshops, and a product showcase.
The keynote address, delivered by Ralph Woerheide, CEO of Metromation, was particularly interesting. In his presentation, titled Data to Make Artificial Intelligence Do Its Work in the Automotive Paint Industry, Woerheide stated that data is the currency that makes a digital ecosystem work. He noted that AI algorithms work effectively because they are trained on datapoints, and that the amount of data needed to train a machine learning algorithm exponentially increases with complexity. But its not all about quantity; quality is central, especially when we speak about data in our industry, where we can run into problems with validity or representation of data.
The paint and coatings industry faces complex interactions of materials and processes. And while some data is shared between partners in the coatings supply chain, he noted that there is not enough of it for statistical viability. Woerheide specified that AI can speak, but it can’t think. More quality data is needed for AI to reach its full potential. Data exchange is needed to achieve this, but companies don’t want to disclose theirs. According to Woerheide, “Data is the currency, and sharing is the deal.”
He said, "I understand why people don’t want to go further in that discussion; there are a lot of things to consider like data formats, data security, and so on. The most important argument is intellectual property. We can set a market value for a patent, or evaluate the value of a formula according to its profit per gallon of sold paint. But a dataset that helps me save money? If you ask me, that dataset has value, and it is time to talk about putting a price tag on it."
Woerheide posed these questions. “How can we determine the value of a data set that brings value to someone else?” and “What if we find a way to share data for training machine learning algorithms but maintain proprietary knowledge for each stakeholder?”
In addition to his work with Metromation, Woerheide is Vice Chairman of the Board for the German association, Smart Paint Factory Alliance (SPFA), which is dedicated to working on these questions. The aim is to help companies overcome the fear of sharing their data, and explore the potential of the AI tool.
Some ideas that the SPFA is working on include:
- Security requirements: How can data exchange be orchestrated and managed for the sake of maximizing data security?
- Intellectual Property: How can stakeholders make sure that their IP is not disclosed to unauthorized parties?
- Data validation: How is it possible to make sure that the data is representative of the physical properties needed for the process(es)?
- Data value: How can we find a way to put a price tag on data?
I was able to meet with Woerheide at the conference, and am happy to report that he will be writing an article on this topic for us to publish in the near future. I’m sure all of us will be interested to learn more, and to see what the SPFA discovers with its research!
