Datawork

Datawork is the suite of labor involved in creating, preparing, and maintaining data used in information systems and artificial intelligence. It encompasses data collection, labeling, annotation, cleaning, transformation, integration, and metadata creation; it also includes data validation, quality assurance, data provenance, and governance.

It spans both professional roles, such as data engineers, data scientists, and data stewards, and informal or crowd work, including microtasks like image or text labeling, transcription, and moderation. Activities occur across the data lifecycle, from data acquisition to storage and processing, to annotation for training machine learning models, and to ongoing curation and versioning.

The field draws attention to labor conditions, compensation, transparency, and bias in data sourcing. Critics emphasize that datawork is essential yet often undervalued and outsourced to low-paid workers, and there are ongoing calls to improve working conditions, obtain informed consent, protect data privacy, and ensure fair compensation, while also considering appropriate limits to automation.

In practice, datawork is closely linked to data governance, data quality management, and data ethics. It is distinct from data analysis or modeling, though it supports those activities by providing high-quality data assets.

See also Data governance, data labeling, data curation, data ethics.