Journal or Book Title
When artificial intelligence participates in design, the notion of attribution–and accompanying systems of rights, credit, royalties, etc.–is brought into question. Without some means of identifying and negotiating the use of the contributions of non-human authors, works produced by algorithmic systems and Als may lack the requirements to be recognized as works of authorship under international laws or in academic institutions. This deficiency could prohibit databases of digital models, algorithms, and toolpaths, for example, from being appropriately accessed by other Als to improve designs and create new ones. Machine learning is most efficient when it has not only access to data but also metadata: histories and networks of associations. This is critical to the analysis of designs, which are almost never singular works but rather built from numerous parts, previous designs, and the work of multiple authors. Thus, establishing provenance–the sources, such as participants and processes, involved in producing or delivering an artifact–will be critical to the development of designs in the future.
Doyle, Shelby and Senske, Nicholas, "Attribution and Artificial Intelligence: Embedding Provenance with Material Metadata" (2018). Architecture Publications. 109.