Thinking science with thinking machines: The multiple realities of basic and applied knowledge in a research border zone.

Some scholars dismiss the distinction between basic and applied science as passé, yet substantive assumptions about this boundary remain obdurate in research policy, popular rhetoric, the sociology and philosophy of science, and, indeed, at the level of bench practice. In this article, I draw on a multiple ontology framework to provide a more stable affirmation of a constructivist position in science and technology studies that cannot be reduced to a matter of competing perspectives on a single reality. The analysis is grounded in ethnographic research in the border zone of Artificial Intelligence science. I translate in-situ moments in which members of neighboring but differently situated labs engage in three distinct repertoires that render the reality of basic and applied science: partitioning, flipping, and collapsing. While the essences of scientific objects are nowhere to be found, the boundary between basic and applied is neither illusion nor mere propaganda. Instead, distinctions among scientific knowledge are made real as a matter of course.