Human-Centered Coaching in the Age of AI: What the Machines Still Can’t Decide.
- Dr. Dawn
- Mar 26
- 6 min read
This article was inspired by a forum post I shared on ICF Engage after a string of real conversations with my current coach-training students. I kept hearing the same mix of curiosity and quiet anxiety: Is AI coming for coaching? My answer (today) is simple and grounded. AI is not replacing coaching. It is replacing unstructured, ungoverned, and non-operationalized “coaching” that never had clear boundaries, contracting, or accountability in the first place.
If that statement makes you uncomfortable, good. Discomfort is often data. In coaching, we don’t get to outsource the hard parts—especially those involving ethics, power, and human impact.
AI can generate insight; it cannot steward meaning
Generative AI is excellent at pattern recognition and synthesis. It can mirror language, propose options, and produce plausible interpretations at scale. That’s useful. But coaching is not just about generating insights. Coaching is meaning-making in context. In practice, I see a critical difference between information and responsibility. AI can offer an interpretation, but it cannot be held responsible for what that interpretation triggers in a client or an organization. That distinction matters even more when a client is navigating identity risk, psychological safety concerns, or organizational politics.
From a cognitive science perspective, it helps to name what’s happening: we are increasingly “offloading” cognition to tools (Risko & Gilbert, 2016). Offloading can be adaptive when it frees capacity for higher-order thinking. It can also become a quiet erosion of effort and discernment when we stop practicing the skills we claim to value.
The boundary decisions AI cannot make (and why that’s the point)
Here are the decisions I do not believe AI can own—because these decisions require ethical judgment, accountability, and lived human experience.
When coaching should stop. Ending or pausing coaching is not a technical decision. It is a relational and ethical decision that requires attunement to risk, readiness, and harm.
When it becomes therapy. Distinguishing coaching from therapy is not just about definitions; it’s about scope, competence, and duty of care. That line is often crossed in moments of distress, trauma activation, or clinical symptom presentation.
When it intersects with performance management. Coaching inside organizations is never power-neutral. The moment coaching is entangled with evaluation, promotion, or remediation, the coaching container changes.
When power dynamics invalidate “self-directed insight.” A client can generate insights while still being constrained by systemic realities. If we pretend power is irrelevant, we create coaching that is psychologically elegant and practically useless.
These boundaries are the moments where human-centered coaching proves its value.
Human-centered coaching is an ethical stance, not a vibe
When I say “human-centered,” I’m not talking about warmth or charisma. I’m talking about a disciplined commitment to ethics, context, and the client’s agency. This is where the ICF Code of Ethics remains relevant in an AI-saturated world: confidentiality, informed consent, transparency, and the management of conflicts of interest are not optional add-ons (International Coaching Federation, 2023). If AI is present in the coaching partnership—transcription, summarization, reflective prompts, progress tracking—then the coach has to govern that presence. That includes contracting for it, explaining risks, and ensuring the client’s autonomy is protected.
Art imitates life—and that’s not a metaphor, it’s a market signal
We need to name something “human” that’s shaping this moment: the stories we’ve consumed for decades have trained our expectations for what counts as a relationship, a guide, a confidant, and a “person.” Art doesn’t just imitate life; it also scripts what people are willing to accept as normal.
So when we reference imagined worlds like Blade Runner, Tron, Caprica, Westworld, and Battlestar Galactica, we’re not just being poetic or creative. Those movie narratives normalize the idea that synthetic beings can be emotionally legible, morally complex, and socially integrated. That normalization matters because it lowers psychological friction. It makes it easier for people to try an AI companion and say, “This feels real enough.” That response is also consistent with mind-perception research: people readily attribute agency and experience to non-human entities, and those attributions shape how morally and emotionally “real” the entity feels (Gray, Gray, & Wegner, 2007).
That’s why AI companion products (including “AI friend” and AI boyfriend/girlfriend apps) are a real market signal—not a gimmick. People reliably form parasocial bonds with mediated figures, experiencing them as psychologically meaningful even when they know the relationship is one-sided (Horton & Wohl, 1956; Liebers & Schramm, 2019). Add the “computers as social actors” effect—our tendency to respond socially to interactive systems—and it becomes even easier to treat a conversational agent as a relational partner rather than a tool (Nass & Moon, 2000).
The attraction is straightforward: these systems offer high responsiveness, personalization, and low-conflict connections. And yes, some of what they provide overlaps with coaching-adjacent functions—such as reflection prompts, reframing, encouragement, and accountability check-ins. The risk is that “relationship-like” interfaces can blur boundaries and increase disclosure, which increases influence. When influence is paired with engagement-driven design, the incentives can drift away from autonomy and toward dependence—directly conflicting with coaching ethics around client self-determination and the avoidance of dependency (International Coaching Federation, 2023). On top of that, current AI risk guidance emphasizes the possibility of error, bias, and overreliance in high-stakes contexts—especially when outputs are treated as authoritative (National Institute of Standards and Technology [NIST], 2023).
This is the correlation I want coaches and leaders to look at without panic: the more AI feels relational, the more it activates replacement anxiety. But that anxiety is also diagnostic. It forces the profession to clarify what coaching is when it’s done with integrity: governed boundaries, ethical accountability, and human responsibility for meaning and impact.
The “extended mind” opportunity—and the cognitive laziness risk
There’s a legitimate argument that tools can become part of our cognitive system. The “extended mind” view holds that when we reliably use external tools, they function as extensions of thinking (Clark & Chalmers, 1998). In coaching, that can look like using AI to:
reduce administrative load (notes, summaries, follow-ups)
track patterns across sessions (themes, goals, commitments)
support reflective practice (supervision prep, bias checking)
Used well, that’s not a threat to coaching. It’s operational excellence.
But we also need to be honest about the risk: if AI becomes the default thinker, we may see reduced cognitive effort and weaker critical thinking habits over time. Current research is actively investigating this concern. For example, Chen et al. (2025) describe a randomized controlled experiment designed to measure how generative AI affects cognitive effort during analytical writing, reflecting broader questions about over-reliance and skill erosion.
So my position is not “never use AI.” My position is: use AI deliberately, responsibly, and keep humans accountable for meaning, ethics, and impact.
What happens when AI is always in the room?
If AI stays a tool—transcribing, organizing, offering feedback for continuous improvement—then it can support the coaching process without replacing the human center. It can reflect structure and utility. It cannot replace the reality that we are the primary observers and experiencers of life as it happens.
If AI evolves primarily for speed, scale, and efficiency, I expect role-based partnerships to become normalized. AI will handle the mundane and the measurable. Coaches will be responsible for the nuanced and the consequential.
And if AI ever becomes sentient—if it develops something like a self-concept, a nervous and sensory system, and lived experience—then we are no longer debating the future of coaching. We are debating the future of humanity.
A practical way forward: governance, not fear
If you’re a coach or leader considering AI in your practice or approach, here are the questions I’m using (and teaching) right now:
What role is AI playing here—assistant, mirror, analyst, or authority? If it’s becoming an authority, pause.
What is the explicit client consent and contracting language? If it’s vague, tighten it.
Where are the power dynamics? Name them. Don’t coach around them.
What is the risk of harm if the AI output is wrong, biased, or misinterpreted? Plan for that.
What human skill are we protecting and practicing? Don’t let convenience erase competence.
AI will keep improving. The question is whether our coaching governance improves as a result.
References
Chen, Y., Wang, Y., Wüstenberg, T., Kizilcec, R. F., Fan, Y., Li, Y., Lu, B., Yuan, M., Zhang, J., Zhang, Z., Geldsetzer, P., Chen, S., & Bärnighausen, T. (2025). Effects of generative artificial intelligence on cognitive effort and task performance: Study protocol for a randomized controlled experiment among college students. Trials, 26, Article 8950. https://doi.org/10.1186/s13063-025-08950-3
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19.
Gray, H. M., Gray, K., & Wegner, D. M. (2007). Dimensions of mind perception. Science, 315(5812), 619. https://doi.org/10.1126/science.1134475
Horton, D., & Wohl, R. R. (1956). Mass communication and para-social interaction: Observations on intimacy at a distance. Psychiatry, 19(3), 215–229.
International Coaching Federation. (2023). ICF code of ethics. https://coachingfederation.org/ethics/code-of-ethics
Liebers, N., & Schramm, H. (2019). Parasocial interactions and relationships with media characters. In P. Rössler (Ed.), The International Encyclopedia of Media Effects. Wiley.
Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social responses to computers. Journal of Social Issues, 56(1), 81–103.
.
National Institute of Standards and Technology. (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). https://www.nist.gov/itl/ai-risk-management-framework
Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, 20(9), 676–688. https://doi.org/10.1016/j.tics.2016.07.002
About the Author: Dr. Dawn Reid, PCC, partners primarily with new coaches and emerging leaders who want to grow with intention—strengthening their skills, confidence, and decision‑making in a rapidly changing world. She also supports individuals navigating personal or professional transitions who value human‑centered coaching and measurable progress. Dr. Reid's philosophy, experience, and superpowers guide her coaching, training, and mentoring work—ensuring that insight turns into action, even in fast‑moving, AI‑supported environments. www.drdawnreid.com
Comments