Defining and assessing AI literacy for researchers across the research lifecycle

Generative artificial intelligence (AI) is now embedded across the research lifecycle, from question formulation through dissemination. These systems create new opportunities for efficiency and access to knowledge. They also introduce significant risks for research integrity, reproducibility, and scholarly agency. Although existing work on AI literacy provides useful foundations, most frameworks focus on students, citizens, or general digital skills and offer limited guidance for researchers as knowledge producers. Drawing on the concept of cultural intermediaries and an established framework of functional, critical, and rhetorical literacy, this paper conceptualizes AI as a cultural intermediary in research. It argues that AI systems mediate scholarly norms, priorities, and forms of expression not through intentional curation but through structural mechanisms such as pattern reproduction, training data composition, and interface design. This extension reframes the cultural intermediary concept from an actor-centered to a function-centered one, and shifts attention from whether AI intends to shape research to whether it produces intermediary effects. Building on this framing and informed by sustained practitioner engagement with researchers across institutional contexts, we propose a capability map that defines AI literacy for researchers across functional, critical, and rhetorical dimensions and maps these capabilities across stages of the research lifecycle. An accompanying assessment rubric specifies observable behavioral indicators at three performance levels. These contributions position AI literacy as a form of scholarly practice rather than a set of technical skills, providing structured guidance for supporting responsible, transparent, and human-centered research in AI-mediated environments.