Chemistry computer science and product development will be increasingly interwoven

How widespread is the use of AI in the paint and coatings industry?

Erik Sapper: My take is that many companies are in the early stages of exploring the implementation and use of artificial intelligence (AI), machine learning (ML), as well as automation in their existing R&D workflows. It’s a tricky task to manage because the new technologies can be so disruptive, and not necessarily in readily apparent, controllable, or beneficial ways. I would characterise the landscape by saying that use is widespread, but optimisation and mastery of using AI and ML tools is far from being the norm.

What are examples of how AI can help in paint formulation?

Sapper: AI is being used across the paint R&D, formulation, and production cycle, from optimizing supply-chain logistics, to using predictive analytics to determine the market demand for new products. Within those two extremes exists the chemistry and formulation that we are so familiar with: AI tools are helping to make sense of synthesis and formulation results, creating synthesis/structure/formulation/property relationships quicker than the standard empirical analysis approaches.

Fed with the appropriate starting data, AI can make predictions for a given synthetic procedure or proposed formulation but can also use the learned models to suggest novel procedures, materials, and formulations. This latter implementation is where chemistry, computer science, and product development will be increasingly interwoven as we become more comfortable with these digitally enhanced workflows.

How complex is it to implement AI/ML workflows in the laboratory?

Sapper: It is a complex task, to be sure. Foremost, an organisation must be supportive of changing its approach to science and product development, from a keep-the-winners mentality to one of all-experiments-create-usable data. This removes pressure on formulators, but also necessitates a clear vision for data management; that is, data creation, control, and sharing across the organization. If culture and data management have a strong foundation in place, in-house expertise in building and deploying AI and ML tools can grow more organically.

Usually, however, there is much data housekeeping that must be performed first, before widespread AI implementation across the organisation; and oftentimes companies may want to partner with outside labs, organizations, or universities to bootstrap and accelerate their efforts.

Event tip

Erik Sapper will also be presenting at the European Coatings Conference Automation. The conference on June 22 and 23 in Stuttgart, Germany, will give impulses for daily work as well as perspectives from the academic environment on digitisation and automation.