Essays.
What I think about AI, experimental design, and where coatings R&D is going.
-
Stop asking how to use AI. Start building.
I think the question "how to use AI" is a bit like asking "how to use electricity". You can use it to turn on a light, to run a motor, to heat your house, to do almost anything. So the better question is what annoys you in your daily work? Or what do you want to build that you alone can't do?
-
Everyone gets their own software
How will we work with computers when AI does most of the work? The answer might be generative UI, where you describe your goal and an AI builds a custom interface just for you, showing only what you need, only when you need it.
-
You no longer need to know DoE to use it
An AI agent can now handle the expertise DoE used to require, choosing the right design, analyzing the results, and picking which experiments to run next. What is left for you is to set the right goals, not to decide which experiments to run.
-
Every experiment is a point in the design space
Every experiment is a point in the design space. Changing one factor at a time covers only the edges. A factorial design covers the whole space with the same handful of runs.
-
Every experiment teaches something. The question is how much.
Every lab runs experiments. What sets two labs apart is how much each one teaches. Four stages separate a lab that repeats itself from one that learns, each teaching more than the last.