A more human way for PhD researchers and supervisors to explore AI, make thoughtful choices, and protect the things research is really for.
Inspired by Janine Arantes's Human (de)Centred AI in Doctoral Research (2026).
A paced, captioned visual prologue you can play, pause or read at your own speed. No synthetic voice required.
The framework moves the conversation beyond permission and productivity. Open each lens to hear the questions hiding beneath the surface.
Where does assistance help? What should remain human-led? When is AI itself something to examine?
Look at the developing researcher, people affected by the work, and infrastructures, resources and Country.
Separate what is methodologically defensible from what is permitted and what requires formal authority.
Six gentle steps adapted from the workbook's decision record. Do this with a supervisor, a colleague, or on your own. Your notes stay in this browser; nothing is submitted to a server.
Start with the research intention, not the tool. There is no perfect answer.
Choose a situation from the workbook and explore a defensible next step. These are illustrative scenarios, not approvals.
A doctoral candidate hopes AI can speed up a scoping review. But suggested citations can be invented or incomplete.
A candidate wants a commercial model to suggest themes from participant interviews. Is removing names enough?
A candidate relies on approved AI reading assistance; a supervisor worries about intellectual independence.
As you work through the conversation, your thinking appears here. Copy or print it for a real supervisory discussion. It is a reflection record, not an ethics or institutional approval.
Please don't enter identifiable participant data, confidential material, or restricted knowledge here. Browser-local notes can still be visible to others with access to your device.