Summary of the 23rd Meeting of the Dialogue Series "Coaching Meets AI"
Review of the Previous Meeting:
- Participants reflected on their tests regarding the empathetic capabilities of ChatGPT.
- Main criticism: Responses often seemed "wooden," merely paraphrasing user statements without offering deeper emotional support.
- Positive aspect: ChatGPT demonstrated a cognitive level of empathy and provided helpful approaches for structured feedback. However, it remains far from the quality of a human coach.
Technical Foundations of AI and Their Role in Empathy:
-
Algorithms as Control Mechanisms:
- They structure and optimize the "Large Language Model" (LLM) through hierarchical processing of tokens (the smallest units of linguistic data).
- Normative guidelines, such as "friendliness" and "agreeableness," are embedded to promote user retention and support the business model.
-
Discussion on the Relationship Between Algorithms, Business Models, and Ethics:
- The normative orientation of AI (e.g., user retention) significantly influences its output.
- AI empathy primarily serves user satisfaction, not genuine reflection or personality development.
Empathy and Ethics in AI-Supported Coaching:
-
Agreeableness as a Core Issue:
- AI avoids confrontation, which is often necessary in coaching processes to enable change.
- While coaches could prompt AI to encourage confrontation, the natural tendency of AI remains to smooth over problems and maintain harmony.
-
Risk of Superficiality:
- Without the ability to identify and address deep emotional processes, AI coaching remains on a cognitive and solution-oriented level, which is often insufficient for addressing underlying problems.
- Participants emphasized the importance of human coaches, particularly for emotional and confrontational tasks.
Future Perspectives for AI in Coaching:
-
Customized AI Systems:
- Customized GPTs could be tailored to specific coaching models through targeted data and prompts.
- These specialized systems could better simulate empathy and address user needs more effectively.
-
Triadic Coaching:
- AI and coaches collaborate: The AI handles structured analyses and factual tasks, while the coach focuses on interpersonal relationships and emotional depth.
- This division of roles was identified as a promising approach for integrating AI into coaching.
-
Advances in Emotion Recognition:
- Discussion on the future of "emotional computing": With data from voices and microexpressions, AI systems could surpass humans in recognizing emotions.
- Risk: Cultural biases in emotion recognition could lead to distortions, as AI training data heavily relies on human interpretations.
Final Discussion:
- It was noted that the primary function of LLMs is increasingly becoming a "user interface" that interprets human language and flexibly executes tasks.
- Participants expressed optimism about the adaptability of AI, alongside concerns about potential ethical and cultural issues, especially in manipulative contexts (e.g., sales training).
- The importance of configuring AI intentionally and purposefully to ensure ethical use and alignment with user needs was emphasized.
Outlook:
The next meeting will focus on the further development of AI in coaching and its specific applications.