AI in coatings science minus the hype.
This two-day workshop gives you an honest picture of what AI can do in a lab today, and how to get started.
Since ChatGPT launched, AI has been everywhere. But what can it actually do for us in coatings science? My honest opinion is that AI is going to change how coatings are developed but getting there means looking past the hype first.
AI has been around for many years. At its core it is statistics, and statistics has been around much longer. What changed with generative AI (ChatGPT and the tools that followed) is that this kind of analysis became accessible to anyone who can type a question. That is worth paying attention to.
Data is the foundation of all of it. Good, high-quality data has always been hard to gather. Every department and every person had their own system for doing and recording things. AI can help fix that. It can handle documentation, structure data, and make design of experiments the default way of running a study rather than the exception.
This two-day course gives you an honest picture of where AI stands today, what it can handle in a coatings lab, and how to get started. You work with real tools on real datasets. You leave knowing what exists and with enough hands-on experience to keep going.
Formulators, lab chemists, and R&D managers in coatings and material science, and anyone in the natural sciences who wants to understand how these tools work and where they fit. No prior knowledge of AI, statistics, or programming is required.
AI, machine learning, and statistics
What these terms mean, how they relate to each other, and where tools like ChatGPT fit in.
Predicting properties
Which approaches exist for predicting properties from data you have, how to build a model, and how to judge whether the result is good enough to act on.
Data as the foundation
What high-quality data looks like and how to generate it.
Guided experimentation
Instead of deciding upfront which experiments to run, a model looks at your results so far and proposes the next experiment most likely to get you closer to your target.
Supporting visual evaluation
AI can build custom tools that make subjective assessment more consistent. Like a delamination calculator or a small webapp that displays reference images next to your result so you can "anchor" a rating.
LLMs in daily work
What LLMs are and how we can use them for patent and literature research, documentation, data analysis, and reporting. How to spot hallucinations and what to do about them.
AI agents
What an AI agent is, what tasks agents can handle today, and where the limits are.
Data safety and GDPR
How can we use all these tools safely?
- Distinguish between AI, machine learning, and classical statistics and decide which approach fits a given task.
- Explain how a large language model works and how to use it reliably.
- Use an LLM tool to draft a test report, search literature, and extract information from various sources.
- Explain what an AI agent is and how it differs from a regular chatbot.
- Identify tasks in your own work that an agent can handle.
- Describe the risks of delegating tasks to an AI agent and the conditions under which it makes sense.
- Build a custom tool that supports a visual test evaluation.
- Describe what high-quality data looks like and how to generate it.
- Use an AI agent to turn unstructured data into a consistent, analysis-ready format.
- Train a model on a prepared dataset to predict properties and judge whether the prediction is trustworthy.
- Explain how model-guided experimentation differs from deciding all experiments upfront and when it is the better choice.
- Identify the data safety risks of a given tool and apply a clear set of rules for what to upload.
- Laptops and datasets are provided and everything is preinstalled. You do not need to install anything, request IT approval, or bring company data. You can experiment freely without any data safety or confidentiality concerns.
- Tool subscriptions for everything used in the course are included and remain active for three months after the course so you can keep practicing.
Want a seat in the next workshop?
There's no workshop scheduled right now. Add your name and you'll get priority for the next one.