There’s no escaping discussions about AI, whether that’s concerns about job losses due to AI, the environmental impact, impressive new innovations, or the fear of it evolving beyond our control. Like many professions, we’re seeing AI used more and more in the coaching space, but what does this mean for us as coaches?
AI can now generate coaching-style questions with impressive ease. It can offer reflective prompts, summarise themes, suggest exercises, help someone explore options and produce language that sounds thoughtful, curious and supportive. Some of that is useful, some surprisingly good, and perhaps makes this kind of interaction more accessible to those who would be unable to work with a human coach.
So the interesting question is no longer whether AI can imitate parts of coaching. It clearly can. The more useful question is what that imitation reveals about the way we understand coaching itself.
If coaching is only the production of open questions, reflections and action steps, then AI is already close to much of the visible work. If, however, coaching is a relational, ethical and meaning-making practice, then the rise of AI exposes the difference between sounding like a coach and actually being one.
Coaching is more than its outer form
Coaching has always been vulnerable to being mistaken for what it looks like from the outside. The coach listens, asks questions, reflects back and helps the client clarify what is happening. But that’s only part of the practice.
New coaches often focus on their questions, reaching for the ‘right’ question to ask. But as a coach becomes more experienced, the work becomes less about selecting the most insightful question and more about learning to dance with the client and understanding what the moment actually needs. The coach learns to listen not only to the content of what is being said, but also to hesitation, contradiction, tone, pace and what appears to be happening in the relationship. Through reflective practice, they become more able to notice their own urge to rescue, interpret or move things on too quickly.
AI is a useful mirror here because it can reproduce much of the language coaches use. It can ask sensible questions about values, assumptions, options and next steps. What it cannot do in the same way is participate in the living relationship between coach and client.
Imagine a client says, “I know I should leave this job, but I can’t seem to do it.” There are many plausible responses. The coach could explore what is keeping them there, what leaving might represent, what feels risky, or why the word “should” matters.
The skill isn’t simply in generating one of those questions. It lies in attending to what the statement carries for this particular person at this particular moment. There may be grief, loyalty, fear, relief, shame or something the client has only just begun to recognise. The same question could be useful in one session and clumsy in another.
Curious language is not the same as curiosity
AI also exposes how easy it is to reproduce the language of curiosity. “Tell me more about that.” “What feels important here?” “What might you be assuming?” These are familiar coaching phrases and, used well, they can be useful. But the words themselves do not create curiosity.
For the human coach, curiosity is a discipline because we bring our own assumptions into the room. We have ideas about success, work, relationships, courage and what change should look like. We might hear a client say they want promotion and assume ambition is central, or hear that they want to slow down and assume burnout is the issue. Sometimes those interpretations will be useful. Sometimes they will narrow the conversation before the client has had a chance to discover what is actually there.
This is where the Unknowing Principle becomes important. Unknowing isn’t simply the absence of giving advice, or leading the client to a realisation. It’s the practice of remaining curious, and of sitting with uncertainty when certainty would be easier.
Human coaches have to work at this because we experience the temptation to know. We may want to be useful or insightful. We may feel uncomfortable when we are not sure where the conversation is going. Staying curious requires us to notice those impulses rather than quietly turning them into direction for the client. That kind of reflexivity is one of the places where the human work becomes visible.
The client is not information to be processed
Technology is good at organising information. A client describes a dilemma, themes can be identified, patterns summarised, options generated and possible actions proposed. Sometimes that is exactly what someone needs.
But a client’s experience is not simply a set of inputs waiting to be arranged more clearly. It’s lived through relationships, history, identity, culture, emotion and the body. It may be contradictory, incomplete or difficult to articulate, and sometimes the client doesn’t yet know what they mean by what they are saying.
Transformative coaching often involves staying close to that not-knowing for longer than feels comfortable, and spending time allowing the client’s lived experience to emerge and be understood. This is part of the phenomenological quality of coaching: paying attention to the client’s experience before rushing to explain it.
AI can help organise language. A coach can also organise language. The distinction is that coaching sometimes requires us to resist organisation for a while because the client’s experience is still forming.
Relationship changes what becomes possible
Perhaps the clearest difference appears in the relationship itself.
The same intervention can land very differently depending on the trust between coach and client. A direct challenge may be useful with someone who knows the coach well and has invited that kind of honesty, while the same challenge may feel unwelcome elsewhere. Silence can create space for one client and feel uncomfortable and exposing to another.
This is why the relationship is so central to coaching. It shapes what can be said and what can be heard.
Trust is built over time through contracting, consistency, attention, boundaries, respect and the experience of being truly heard. Within that relationship, a client may risk expressing something they have not yet fully admitted to themselves, or allow an idea to remain unfinished rather than presenting the polished version. AI can produce empathic language, and that language can sometimes be genuinely useful or comforting. But empathic wording and relational trust are not the same thing.
Ethics is more than following a protocol
The use of AI for coaching raises practical questions about confidentiality, consent, data and transparency, all of which matter. But ethical coaching also involves ongoing judgement about the nature of the work itself.
A coach may need to recognise when something is moving beyond their competence, when the coaching contract needs revisiting, when a client is becoming unusually dependent on the relationship, or when the coach is colluding with a story that needs closer examination. They may need to think about competing responsibilities when coaching is sponsored by an organisation, or notice that their own desire to be helpful is beginning to shape the session.
A practitioner has to take responsibility for how they show up, what they notice, what they fail to notice and what they choose to do next. That accountability is part of coaching, but less apparent when AI is used as a coach. We’ve not yet reached the stage where AI is engaging in reflective practice with a Coaching Supervisor.
AI may help coaching define itself more clearly
If AI can generate strong reflective questions, then questions cannot be the defining feature of professional coaching. If it can summarise patterns and suggest action steps, then neither can those things. This may be useful pressure on the profession because it pushes us beyond a toolkit understanding of coaching.
It asks us to pay more attention to the qualities that have always sat underneath technique: presence, relational judgement, ethical responsibility, self-awareness, humility and the capacity to work with uncertainty.
For transformative coaching, that feels especially relevant. The work is not only concerned with helping someone solve a problem or reach an outcome. It is concerned with how they understand themselves, how they make meaning and how their relationships, assumptions and ways of seeing shape the choices available to them.
Perhaps the best approach is not to defend human coaching by pretending the technology is incapable of anything valuable. Instead, we can communicate the difference between reproducing the process of coaching, and actually practising the deeper craft.
What remains distinctly human may not be the ability to ask an excellent question. It may be the intuition to know when not to ask it, the humility to remain uncertain about what the client means, the responsibility to notice what is happening in the relationship, and the capacity to stay present and curious when the conversation cannot yet be reduced to a question and answer.
As AI becomes better at sounding like a coach, that distinction may become even more important.
