The results: How to prompt

The first step defines the role or persona that the AI should adopt. This helps the system frame its response from a specific perspective.

For example, instead of writing:

“Make a lesson plan for Dutch A2 learners”,

a teacher can specify:

“Act as an experienced teacher of Dutch as a second language for adult learners with 20 years of experience in effective pedagogy.”

Defining the role:

• improves relevance of the output;

• reduces generic responses;

• aligns AI output with professional language and practice.

The role does not describe who you are as a teacher, but who the AI should pretend to be

The task clearly states what kind of output you expect. Vague tasks lead to vague results.

Examples of well-defined tasks include:

• develop a lesson plan;

• reformulate an existing activity;

• suggest differentiation options;

• review a lesson plan critically.

For adult L2 teachers, it is often helpful to specify the mandatory components of the task (e.g. lesson objectives, warm-up, main activities).

Contextual information situates the task in a real teaching situation. This includes elements such as:

• type of course (integration, workplace language, general L2);

• teaching format (face-to-face, online, hybrid);

• lesson duration;

• available infrastructure.

Context helps the AI avoid unrealistic suggestions and increases practical usability.

Adult second language classrooms are rarely homogeneous. The target group step allows teachers to describe learner characteristics such as:

• CEFR level;

• age range;

• literacy level;

• learning needs or constraints.

This step is crucial for differentiation and inclusion. Even a short description can significantly improve output quality.

Parameters define boundaries and formats. They prevent AI output from becoming too long, too complex or pedagogically unfocused.

Examples include:

• lesson duration;

• number of activities;

• inclusion of listening or speaking tasks;

• use of specific frameworks (CEFR, Bloom’s taxonomy).

Teachers in the GLOW project reported that clear parameters were one of the strongest predictors of usable output.

Tone of voice determines style and register. In lesson planning, this often concerns whether output should be:

• formal or informal;

• concise or elaborated;

• instructive or reflective.

Specifying tone helps ensure that AI output fits the teacher’s professional context and personal style.

The final step invites interaction. Teachers can ask the AI to:

• ask clarifying questions;

• wait for feedback before finalising output;

• revise its own response.

This step reinforces the human-in-the-loop principle and turns prompting into a dialogue rather than a one-off request.