Instrument Or Power: How Academic Abstracts Represent Artificial Intelligence In Intelligence And Police Work
Year: 2026
Type: preprint
Abstract: Artificial intelligence (AI) is entering intelligence work, including the criminal intelligence that supports police investigations and the anticipatory intelligence that serves prevention and warning, and the way academic writing pictures it may orient which of its risks receive attention. This paper asks how 142 academic abstracts on AI in intelligence work represent AI, through the theory of social representations. A Reinert lexical classification, which sorts passages of text by the words they share, divides them into three classes, read through anchoring and objectification. AI appears either as an instrument, built, tested and handed to the analyst, or as a power of states, rival states and their agencies, and power either as something to contain by law and oversight or as a force that transforms. The words of law, oversight and ethics gather with power and are among the least associated with the instrument, so no class presents the instrument as something to contain. A second coding of the abstracts finds that 23 raise the checking of AI output by its users, 12 of them from 2026, and the classification places most of these statements with the instrument: the theme is present, but it is written in the words of the tool, not of control. Abstracts on crime, policing or terrorism raise it more often than the rest (8 of 30). The three-class solution passes the stability rule only under a reading settled after the values were known, and the results describe a purposive sample of abstracts, not the field.
Source: CrimRxiv
Author: Filippo Tansini
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225988527
