From Toil to Thought: Designing for Strategic Exploration and Responsible AI in Systematic Literature Reviews

AI transparency is essential for scholarly trust: The paper found that researchers need to understand why AI systems make screening decisions about research papers. ARC's approach of providing "transparent AI reasoning" rather than black-box automation helps researchers verify AI judgments and maintain scholarly agency—a key principle for responsible AI implementation in knowledge work.

Automation should reduce busywork, not replace expertise: The study identified a critical tension between automation and scholarly agency. Rather than fully automating literature review decisions, the responsible approach is using AI to eliminate administrative overhead (managing multiple databases, query refinement) so experts can focus on higher-level strategic thinking and judgment calls that require human expertise.

Tool fragmentation creates hidden costs: The research with 20 experienced scholars revealed that scattered, disconnected tools impose significant "cognitive load" on researchers. This friction doesn't just slow work down—it actively suppresses the exploratory, iterative nature of good scholarship. Responsible AI design should consider how systems integrate into existing workflows rather than adding more fragmentation.

Explainability enables long-term accountability: The paper emphasizes that AI systems should support "verifiable judgment" and "long-term maintenance of knowledge synthesis." This means AI decisions in literature reviews need to remain auditable and understandable over time, not just at the moment of screening—important for maintaining scientific integrity and reproducibility.