Reproduce “Data-Centric Artificial Intelligence” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Johannes Jakubik, Michael Vössing, Niklas Kühl, Jannis Walk, Gerhard Satzger

Researchers from KIT and the University of Bayreuth introduce data-centric AI as a paradigm complementing model-centric AI, providing a structured framework for data engineering activities like refinement and extension. The work outlines a research agenda for the Business and Information Systems Engineering community to advance robust AI systems through systematic data quality improvements and management.

Key insights

  • Model-centric and data-centric AI are complementary paradigms; a holistic approach to AI development requires integrating both to achieve effective and efficient systems.
  • Data quality in data-centric AI is evaluated based on its impact on machine learning model performance metrics, rather than solely traditional data quality measures.
  • The Business and Information Systems Engineering (BISE) community is uniquely positioned to contribute to data-centric AI through expertise in organizational processes, human-in-the-loop systems, and data governance.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.