Why AI-written references can’t be trusted. General AI chatbots are known for writing references that look perfect but don’t exist: plausible authors, a real journal, a believable title and a DOI that leads nowhere. Students and staff have been caught out by them. The safe approach is the one this tool takes. Search real databases first, then let the AI help you read and organise what was actually found, and never let it write a reference itself.
Start from a focused question. Good searching starts with a good question. Frameworks such as PICO (population, intervention, comparison, outcome) for clinical questions, or PEO (population, exposure, outcome) for qualitative ones, help you name the parts of your topic. A focused question finds papers you can use. A broad one finds thousands you can’t.
Know your study types. Not all evidence is equal. Systematic reviews and meta-analyses pull together many studies using an explicit method, so they are usually the best place to start. Randomised controlled trials test whether something works. Cohort and cross-sectional studies describe patterns. Qualitative studies explain experiences and meanings. Narrative reviews and expert opinion are useful for orientation but carry the most bias. Match the type of study to the question you’re asking.
Read beyond the abstract. An abstract is the authors’ own summary, written to be read. It leaves out limitations, small samples and awkward results. Use summaries and abstracts to decide what to read, then read the methods and discussion of the papers you rely on. That matters most for anything you’ll teach as fact or put in a curriculum document.
Check currency, but don’t ignore the classics. In fast-moving fields such as clinical practice, guidelines and technology, recent evidence matters, and a five- or ten-year window is sensible. In others, a landmark paper from twenty years ago may still be the one everyone cites. Use the date filter deliberately, not by habit.
Give students free full text where you can. Open-access papers can be read anywhere, without a library login. For required readings, prefer open-access versions or check that your library subscribes. The tool flags open-access papers and links to the free version.
Teach referencing as part of academic integrity. Show students where the references came from and how to check them: open the DOI, find the paper and confirm the details. Ask them to verify a citation in class. It’s a quick lesson in why referencing matters, and a useful habit in an age of AI-generated text.
Use referenced writing as a model, not a shortcut. A referenced article written from a set of sources makes a good model of synthesis. It compares and contrasts studies rather than summarising them one at a time. Use it as background reading, a starting point for your own writing, or an example to critique with students. Because it’s written from abstracts, check its claims against the full papers before you publish or share it. If students submit work, follow your institution’s policy on AI use.
Follow your institution’s style guide. APA 7, Harvard and Vancouver all have local variations, especially Harvard. The formatting here follows the standard rules (Harvard follows Cite Them Right), but your library’s guide has the final say. Import the RIS or BibTeX file into a reference manager to switch styles and keep your library tidy.