17. Practical Workflows and Checklists
These workflows turn general principles into repeatable practice.
17.1 Research project workflow
Start by writing the research question and inclusion criteria yourself. Use a conversational assistant to refine terms and identify possible concepts. Search scholarly databases, record queries, and retrieve original papers. Use research tools to organise candidate evidence, then read and verify the sources. Draft the synthesis in your own analytical structure. Use a writing assistant for clarity and consistency, then verify every citation and check the institution’s AI policy.
17.2 Dissertation chapter review
Provide the chapter and the approved research questions. Ask for a diagnostic review of alignment, argument, evidence, methodological consistency, limitations, and APA 7 issues. Request exact section references and a prioritised revision plan. Check the critique against the dissertation and supervisor guidance. Make revisions yourself, preserve a version history, and verify all references.
17.3 Business report workflow
Define the decision the report must support. Gather verified company and market evidence. Ask AI to create an outline and list missing evidence. Draft each section with clear separation between sourced facts, assumptions, analysis, and recommendations. Independently verify numbers and external claims. Ask a second reviewer to challenge the recommendations before final approval.
17.4 Data analysis workflow
Describe the research design, variable definitions, sample, and intended analysis. Ask AI for a plan and code draft. Review the code and test it on a copy of the data. Compare outputs with the statistical software and check assumptions. Have a knowledgeable reviewer examine the interpretation, especially causal language and generalisability. Keep an auditable record of transformations and decisions.
17.5 Final AI-assisted work checklist
- Does the output answer the actual question and fit the intended audience?
- Have all facts, quotations, calculations, references, and links been checked?
- Are uncertainty, assumptions, and limitations stated accurately?
- Have confidential or personal data been protected?
- Does the output comply with institutional, professional, legal, and publisher rules?
- Have bias, accessibility, cultural relevance, and possible harm been considered?
- Can the author explain and defend the final content?
- Has AI use been disclosed where required?
- Is there a human owner accountable for the result?