Forcing generalization: technical art as (synthetic) data work

James Steinhoff
Additional contact information
James Steinhoff: University College Dublin

No t7kvz_v1, MediArXiv from Center for Open Science

Abstract: Synthetic data has recently been proposed as an alternate means of procuring data for training AI which dispenses with data work. However the labour required to produce it has not been studied. This paper does so by looking at the technical artist: a hybrid programmer and 3D artist recently brought into the AI industry from the games and film industry. I argue that technical art, in the synthetic data context, is data work but of an unfamiliar kind. I demonstrate this through a labour process analysis of procedural asset creation. I show that in the synthetic data context, technical art is governed by the goal of forcing generalization. I suggest that the concept of data work should not ossify to capture only its present state of collection and cleaning, but that a more mutable concept is necessary to track changes in the AI industry. While claims of data work’s coming disappearance are implausible, it seems unwise to overstate its permanency in its present state.

Date: 2026-05-23
References: Add references at CitEc
Citations:

Downloads: (external link)
https://osf.io/download/6a0d8490acbef6a10d8ecaac/

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:osf:mediar:t7kvz_v1

DOI: 10.31219/osf.io/t7kvz_v1

Access Statistics for this paper

More papers in MediArXiv from Center for Open Science
Bibliographic data for series maintained by OSF ().