Why machine-readable data is not just a good idea for AI

New Podcast by Philip Strömert and Steffen Neumann

NFDI4Chem has contributed to the NFDI podcast with a very important topic for chemistry: Artificial intelligence is changing chemical research – but it is only as good as the data it is fed with.

The podcast can be heard on the following platforms:

Artificial intelligence is changing chemical research – but it is only as good as the data it is fed with. In this episode, Philipp Strömert and Steffen Neumann talk about why metadata, ontologies and digital repositories are essential for machine-readable data. How can AI models be trained with retrospective data sets? What needs to change in teaching, laboratory and publication practice? And why is it high time to secure the digital sovereignty of chemistry in Europe?

Topics of the episode:

  • Data management as scientific practice: Why the digital documentation of laboratory processes is not an “extra effort”, but part of good scientific practice – and why you are the first to benefit from it.
  • Metadata: Why data without context is worth little – and how ontologies help to define technical terms uniformly and unambiguously worldwide.
  • ELNs, repositories & annotation: How tools such as NFDI4Chem’s Chemotion ELN and repository already enable machine-readable, FAIR data – directly from the lab and without copy-paste from PDFs.
  • AI needs clean training data: How machine learning makes old data sets usable – provided they are correctly annotated, standardised and findable. Why this is the basis for the AI applications of tomorrow.
  • Securing digital sovereignty in Europe: What is at stake if Europe leaves the standards for chemical research data to others. Why there is no alternative to common, open data infrastructures – for science, business and democracy.

Guests:

Dr Steffen Neumann is a bioinformatician at the Leibniz Institute of Plant Biochemistry in Halle (Saale). He develops digital solutions for processing chemical and biochemical research data and is committed to machine-readable, interoperable data sets.

Philipp Strömert is an expert for metadata and ontologies at NFDI4Chem (Task Area 6 “Synergies”) and works at the TIB – Leibniz Information Centre for Science and Technology in Hanover.