Use AI for research in five stages: scope your question, find candidate sources with an AI research tool, read the original sources yourself, use AI to summarise and compare them, then write and cite from the originals. Never cite a source you have not opened, and always disclose AI use where your institution requires it.
What AI is good and bad at in research
AI can save hours on the slow parts of research: narrowing a question, finding starting points, summarising long papers and spotting gaps. But it is unreliable on precise facts and citations, and it cannot replace reading sources yourself. A good workflow uses AI where it is strong and keeps you in charge where it is weak.
Stage 1: Scope the question
A vague question leads to vague research. Use AI to sharpen it.
"My topic is the impact of social media on teenagers. Suggest five narrower research questions suitable for a 3,000-word A-Level project, and explain which is easiest to research with available evidence."
Choose a question you find interesting and that has accessible sources.
Stage 2: Find candidate sources
Use AI research tools with citations, Google Scholar, your library database and reputable organisations. Ask for different perspectives:
"Find recent peer-reviewed studies and official reports on this question, including ones that disagree with each other. Give links."
Rule: treat every source as unconfirmed until you have opened it. AI tools sometimes produce citations that do not exist or misstate what a source says.
Stage 3: Read the originals
Open each source. Check the author, date, publisher and methods. Read the abstract, introduction and conclusion first, then the sections that matter. Take your own notes.
Stage 4: Use AI to summarise and compare
Now AI becomes very useful, working from material you provide:
"Here are my notes on four studies. Create a comparison table: study, sample, method, main finding, limitations. Then identify where they agree and disagree."
"Summarise this paper's method and limitations in plain English. Quote the relevant sentences."
Grounding the AI in your notes keeps it honest. See our prompt patterns for more.
Stage 5: Write and cite from the originals
Write the argument yourself. Use AI for feedback on structure and clarity, not to generate your text. Cite the original sources in your required referencing style; our guide to referencing correctly helps. If your institution requires it, include a short statement explaining how you used AI.
Red flags
- A source you cannot find through a normal search.
- A quote that does not appear in the source.
- Statistics without an identifiable origin.
- A summary that sounds more certain than the original paper.
A sample AI use statement
"I used an AI assistant to help narrow my research question, to suggest initial sources (all of which I located and read myself), and to produce comparison tables from my own notes. All analysis and writing are my own."
Always follow your school or university's specific rules.
Frequently asked questions
Can I cite an AI chatbot as a source?
Generally, no. Cite the original sources. Some style guides explain how to acknowledge AI tools, but they are not evidence.
Is using AI for research allowed at university?
Policies vary widely by institution and even by module. Check before using AI on assessed work.
Which AI tools are best for research?
Tools that show citations and let you upload documents are most useful, combined with academic databases.
