The term "data-centric AI" is particularly at home in the fields of artificial intelligence, big data, smart data and digital transformation. With data-centric AI, the focus is no longer just on the model or the algorithm, but above all on the quality and selection of the data used to train an artificial intelligence (AI).
Whereas in the past, the main aim was to make models smarter and smarter, Data Centric AI focuses on collecting and processing the right, clean and relevant data. This often leads to better results because an AI without good data cannot work effectively even with the best algorithm.
A simple example: Imagine you want to use an AI that automatically recognises invoices from emails and assigns them correctly. With Data Centric AI, you first check whether your data, i.e. the collected invoices, is complete, correct and well labelled. You remove incorrect or duplicate entries and add missing information. Only then is the AI trained.
Data Centric AI increases the reliability and informative value of AI systems - and makes them better because they work with high-quality data from the outset.