Lecture
From Data Challenges to AI Use: What R&D Laboratories Really Need
- at -
- B2.137
- Type: Lecture
Lecture description
Artificial intelligence (AI) is widely discussed in R&D laboratories, yet its effective use depends on two prerequisites: suitable AI approaches and a consistent, usable data basis. In practice, formulation, process, and instrument data are often fragmented, limiting systematic reuse and data-driven decision-making. A Material Intelligence framework addresses this challenge by integrating laboratory data, workflows, and AI into a coherent structure. Within this framework, an agentic AI-based co-engineer supports R&D work by enabling contextual search, document access, and data analysis across structured laboratory data. This allows existing knowledge and experimental results to be made accessible and interpretable early in the development process. Building on this harmonised data basis, machine learning and Bayesian optimisation can be applied for predictive modelling and targeted experimental planning. These approaches support formulation optimisation also under sparse-data conditions. The presentation discusses how Material Intelligence enables the practical application of AI in R&D laboratories and supports efficient decision-making in complex formulation environments.
