MRB uses AI to help people prepare and check research documents. It is assistance, not a decision-maker.
What the AI can help with
- Completeness. Flagging sections or documents that appear to be missing from a submission.
- Consistency. Pointing out where one section seems to disagree with another, for example a sample size that differs between the methodology and the statistics plan.
- Questions for reviewers. Suggesting methodological points a reviewer may want to look at.
Each suggestion refers back to the section it came from, so a reviewer can see why it was raised.
What the AI does not do
It does not approve or reject research. It does not replace a guide, a scientific reviewer or an Ethics Committee. Scientific and ethical decisions remain with authorised human reviewers and committees.
How reviewers stay in control
Every AI suggestion can be accepted, edited or rejected. What the reviewer chose is recorded with the review, so the trail shows a human decision, not a machine output.
Why it is designed this way
Review quality depends on judgement, context and accountability. AI is useful for the checks that are tedious and easy to miss. The decision, and the responsibility for it, belongs to people.