AI can write an answer. It can also frame the question, select sources and build an argument. What does that leave the student to learn?
EAI examines this at each stage of learning. Start with the goal and the action the student needs to make sense of. Then consider AI's role.
Suppose a student is learning to recognise differences in perspective. AI could compare the two historical sources for them. Whether that helps depends on the stage of learning.
Choose a moment. The sources stay the same; what the student needs to learn changes.
The teacher shows how a source's origin and wording relate to its perspective. AI can provide a worked example, checked by the teacher.
Following the reasoning and making sense of it. This does not yet require a piece of writing of their own.
Now the student needs to decide which differences matter and support that choice. A complete AI comparison may have done that work already.
Explaining an unfamiliar word, provided that the word choice itself is not what the student needs to investigate.
AI can point out an unsupported claim. The student then looks for evidence in the sources and revises their reasoning.
Can the student do this without the same help, later on and with different sources?
Take one lesson or task. What someone needs to learn determines which AI assistance fits, not the other way around.
What should the student learn, and how do they get there?
Where are they in that learning process?
Which step must they make sense of themselves?
What does AI do at this particular point?
That agreement does not yet specify which help fits this student and this task. EAI starts with that question.
In his writing, Hans Visser examines what happens before the final result: who notices something, chooses an approach or forms a judgement? The four questions make those decisions more specific.