AI Tools for #Literature - How to use #ChatGPT and #Elicit for Literature Search and Writing

AI-assisted literature search can speed up the early stages of a literature review, but it can’t replace reading, critical judgment, or the work of integrating sources into a coherent argument. The core workflow presented pairs ChatGPT for drafting and argument development with Elicit for evidence gathering—using each tool for what it does best.

The session begins with a caution: AI tools can summarize and surface relevant papers, yet they won’t tell researchers how to place specific evidence in the right part of a thesis, how to connect findings across sections, or how to write with academic rigor. Reading remains essential because it’s the only way to understand how studies relate, evaluate quality, and decide what belongs in an introduction, literature review, or argument chain.

To demonstrate the approach, the example research topic is servant leadership and its influence on environmental behavior, framed within leadership-focused journals. A researcher first asks ChatGPT for the “value/importance of servant leadership for modern organizations,” then repeats the question in Elicit. Elicit returns a more research-oriented output: it aggregates referenced papers, provides summaries (including abstracts and measured outcomes), and can flag missing bibliographic details such as study type or funding source—signals that help decide whether a paper is trustworthy.

Elicit also supports deeper steps needed for a systematic or structured review. The workflow includes asking follow-up questions about a specific paper, requesting PDFs when available, and extracting practical metadata such as interventions, outcomes, and participant counts. Beyond summarizing individual studies, Elicit can be used to check whether a concept has already been studied—helpful when a researcher initially believes there is “no research” on a given relationship. In the example, searching for servant leadership and environmental behavior yields results and even journal ranking information, which helps with assessing where the evidence comes from.

The session then shows how to combine tools for stronger writing. ChatGPT can generate argument content—such as how servant leadership might increase employee engagement and motivation—but it often lacks robust referencing. The workaround is to use Elicit to find supporting citations for each claim. For instance, after drafting an argument about engagement and motivation, the researcher queries Elicit for references that connect servant leadership to those outcomes. The same pattern is applied to environmental behavior: ChatGPT helps articulate mechanisms (e.g., collaboration, empowerment, sustainability-oriented teamwork), while Elicit supplies the papers to substantiate those mechanisms.

Overall, the method is a practical division of labor: Elicit for locating, summarizing, and retrieving research evidence; ChatGPT for reconceptualizing text, developing argument phrasing, and building the narrative structure. The payoff is a faster, more informed literature review—so long as the final step still depends on reading the original studies and integrating them critically into the thesis.