This was our first AI developers work group in 2025! \ud83c\udf89<\/p>\n
A big thank you to @Dr. Johannes P. D\u00fcrholt from Evonik, who gave us an exciting presentation on April 7, 2025, in which he explained key aspects of his research. \ud83d\udc4f
\nExperimentation is the heart of chemistry — can artificial intelligence support people in this? Dr. Johannes Peter D\u00fcrholt answered this question with a resounding yes at this year’s first AI developer working group! <\/p>\n
In his presentation entitled \u201cHow to plan experiments in the most efficient way using probabilistic machine learning?\u201d, the experienced data scientist used an example from chemical process engineering to illustrate how modern methods of statistical experiment planning use machine learning models to predict desired target variables. This enables users to carry out expensive experiments only in the most promising and informative areas of the search space and to make optimal use of the knowledge gained. Stay up to date – and if you are interested in joining our AI developer working group, please register for the next meeting.<\/p>\n
\nHe recommended the \u201cBoFire<\/a>\u201d framework to listeners who are keen to experiment. This is also used within Evonik and is being significantly co-developed by the Ruhr-based specialty chemicals company. <\/p>\n