Meeting Minutes of the fifth meeting of "Coaching meets AI – Foundations, Opportunities, and Limits".
© Prof. Dr. Harald Geißler
The topic of the fifth meeting of "Coaching meets AI – Foundations, Opportunities, and Limits" was: How can coaches and coachees professionally use AI apps like chatGPT in their online dialogue?
It became clear that for answering this question, prompting is of central importance and consists of two fundamental parts:
· Context information about the existing initial or framework conditions
· and processing tasks that the AI should perform with reference to the entered context information.
These two fundamental parts can be subdivided as follows:
· The context information about the existing initial or framework conditions o can be communicated to the AI by the coach after the coach and coachee have together considered and discussed which information is important.
o Additionally, certain text documents can be uploaded or copied to provide the AI with the necessary context information.
o Finally, there is also the possibility that the coach conducts an interview with the client and uploads the transcript of this interview.
• The processing tasks that the AI receives with reference to these context information can consist of o the AI checking with reference to a specific question which context information might still be missing and should be provided o the AI processing the entered context information textually, e.g., creating a report that has a specific structure, adopts a specific perspective, or presents the entered information in a specific form, such as a mind map.
o the AI offering specific tips or advice regarding the problem described by the entered context information, o the AI creating a solution design, such as a project plan or a training concept, regarding the challenge described by the entered context information,
o and the AI analyzing the entered context information with reference to specific criteria.
For these two structured main parts of the prompts, it is true that
• first, the quality of the outcome that the AI delivers based on the context information and the formulation of the processing task is better the more comprehensive and specific both the context information and the work orders are formulated
• second, both the inputs of the context information and the formulations of the processing tasks must be collaboratively developed in the coach-coachee dialogue
• and third, the work results delivered by the AI must also be examined in the coach-coachee dialogue and collaboratively revised so that the coachees can unconditionally make them their own.
Finally, two special prompts are to be mentioned:
• for a task processing that is qualitatively unsatisfactory despite differentiated context information and a precise formulation of the task,
• to admonish or motivate the AI to reprocess this task and put a bit more effort into it
• or to ask the AI to justify the communicated work result with reference to the entered context information or specified criteria.
Here is the link to the video recording of the fifth meeting: https://drive.google.com/drive/folders/1iNtNwHHBAS63rON2JZYvwf7z8BrhGndK?usp=drive_link