You want to prepare a short conversation exercise and start inventing a character. A few paragraphs later, there is already a long biography, several problems and many possible turns. The story looks rich, yet it is unclear what the learner should do or which information the person playing the role may use. Before adding detail, you need a scene that can be explained and rehearsed.
For psychologists and psychotherapists, developing a training scenario also means defining its boundaries. Here we use a non-clinical dialogue with entirely fictional characters and no material taken from patients. The proposal concerns preparing the scene. It is not a treatment protocol, evidence that simulation is effective, or an assessment of professional competence.
Choose a bounded situation
Begin with the task already selected for the exercise. Suppose the aim is to clarify a vague expression before offering a solution. The scene can be simple: an introductory meeting about an optional professional development workshop. One character is unsure whether the workshop will be “useful”, and the other tries to understand what that means.
There is no need to turn this uncertainty into a clinical history. The proposed task requires a reason for meeting, an opening sentence and a few details consistent with it. State what falls outside the scene too: diagnoses, personal assessments and treatment recommendations are not part of this exercise.
The ASPE standards published in 2017 connect scenario materials with objectives, roles and activity preparation. They concern simulation with human role players and provide a design reference here. They do not validate this example or make an AI conversation equivalent to that educational process.
Invent the character from scratch
In our original example, Marta is an imaginary adult deciding whether to attend the workshop. She has read a general activity description and wants to know whether there will be practical exercises. She is also unsure how much preparation time she would need. These details were created for this scene, rather than drawn from the biography of someone encountered at work.
Her opening sentence is: “I don’t know whether this workshop will be useful for me.” The person playing Marta knows that “useful” mainly means being able to try a small activity and understand the preparation involved. The person leading the conversation receives only the general setting and opening sentence. This distribution of information creates the practice task.
If you notice that you have reused a recognisable conversation, set it aside and rebuild the scene from the abstract learning aim. The approach proposed here does not involve changing a real person’s name. Records, transcripts, family details and clinical events are unnecessary for making this exchange concrete.
Give each role the necessary information
Prepare two separate instructions. For the person leading: “You are introducing an optional workshop. Try to understand what the other person means by useful, without deciding for her whether to enrol.” For the person playing Marta: “Your question concerns practical exercises and preparation time. Respond about these points when they are explored, without introducing other problems.”
These sentences are an original working suggestion, not a required script. They distinguish what the character knows from what the other person still needs to discover. If both participants receive the complete explanation of “useful” beforehand, the opportunity for clarification changes. Decide whether that distribution suits your exercise.
Any observer also needs a clear role. In this exercise, the observer can record which information was provided initially and which emerged during the exchange. They should not invent a hidden history or assess Marta’s personality. The task concerns the constructed dialogue, rather than supposed characteristics of the people taking part.
Decide how to handle missing details
A short scenario inevitably leaves some information unwritten. You can distinguish fixed details, wording that may vary and information that must not be added. In our example, Marta’s question and the workshop’s purpose remain stable. Responses can be phrased differently. Diagnoses, crises and personal events are not added to make the scene more dramatic.
If someone asks about something the script does not cover, the role player can explain that the detail is undefined and request a pause. This is an organisational choice for our exercise, not a response that a real person should give. The group can then decide whether to add the information or acknowledge that the question moves beyond the task.
Avoid making a single invented characteristic automatically explain behaviour. An age or occupation alone is not a reason to portray a character as distrustful or uncooperative. Such details can often be removed if the scene does not need them. The aim is to make a situation understandable, rather than represent an entire social group through one character.
Review additions suggested by AI
If you use AI to prepare a draft, ask it to work only with the invented situation and information you have defined. You might request two possible openings or a check for contradictions. Before using the text, compare each proposal with the boundaries. A new difficulty, label or psychological explanation does not become necessary simply because it makes the story read more smoothly.
Section 2.6 of the NIST Generative AI Profile includes risks involving stereotyped content. It is general guidance from 2024, not an assessment of PsyLab. For your material, examine which characteristics the text connects to behaviour, and remove connections you have no reason to introduce.
Keep the initial instructions beside the revised scenario. A written request does not guarantee that AI will follow it. If a proposal changes the reason for meeting or introduces hidden information, revise it before rehearsal. Do not ask learners to adjust to a script that keeps changing while they are trying to practise a defined task.
Rehearse the scene before reuse
Read both instructions with a colleague or trainer and try a short opening. Check whether each person understands their role, whether the question can be explored and whether information emerges without contradiction. If something does not work, change one point at a time and record the reason. A preliminary rehearsal does not validate the scenario; it helps identify parts that need clarification.
Agree how to stop the exercise and how to distinguish the role from the person portraying it. Afterwards, return to the task and the writing choices. To explore an AI workspace that also includes a training Simulator, you can try PsyLab for free. Availability of the module does not guarantee that every scene will be followed. Always review the material produced and retain responsibility for preparation.