The work of the ninth session of the "Coaching meets AI" dialogue series, which took place yesterday, continued to focus on the development of a prompt script (i.e., a sequence of relevant prompts) for a triadic AI coaching that leverages the capabilities of GPT (without text modules).
In doing so, we recognized that each prompt should consist of the following four parts: • A directive on what the coach should communicate to the coachee and how it should be documented • Implementation of the coach-coachee communication • Creation of a protocol of this communication • Instruction to the AI on how this protocol should be processed
Having limited ourselves to the first coaching phase, i.e., problem and goal description, in the session on May 30th, we additionally looked at the second phase, i.e., problem analysis, yesterday and developed the following prompt design, which we tested to some extent:
Analyze the information currently available under the question of what the central causes of my problem are.
After GPT has executed this task, the following follow-up prompts are available, which can be repeated multiple times and combined in different ways as needed: <Coach gives the coachee space to engage with the GPT response> Your analysis has prompted or triggered the following in me: [Coach documents]
Please explain a bit more precisely what you mean by the cause “???”.
<Coach gives the coachee space to engage with the GPT response> Your analysis has prompted or triggered the following in me: [Coach documents]
What causes might lie behind the cause you mentioned, “???”.
<Coach gives the coachee space to engage with the GPT response> Your analysis has prompted or triggered the following in me: [Coach documents]
Create a ranking of the most important reasons that cause my problem.