We live in a fascinating time, when artificial intelligence (AI) seems able to answer almost any question we ask. It is very tempting to use this convenient tool as a personal adviser or a growth assistant as well. In the coaching world, AI is often discussed as a threat to the profession. But when you look at it through the lens of real transformation, a nuance appears that is worth thinking about, and it concerns everyone who asks AI for advice, coaches included.

A mirror that tells you what you want to hear
AI assistants tend to give you comfort. They are inclined to agree, to confirm and to avoid “conflict”. Researchers call this sycophancy. A study by Mrinank Sharma and colleagues, “Towards Understanding Sycophancy in Language Models” (2023), found this behaviour in several leading AI assistants and concluded that it is “likely driven in part by human preference judgments favoring sycophantic responses”. In other words, these models learn partly from people’s ratings of their answers, and people tend to prefer answers that match their own views. Developers are working to reduce this tendency, but it has not gone away completely.

So whether you doubt yourself, are overconfident, or lack the skill to see yourself from the outside, there is a risk that AI will show it all “in a positive light” and strengthen your traits, your way of thinking or your behaviour.

This creates an illusion of progress. If you bring a flawed strategy to the conversation, the assistant may well confirm it with confidence. If you bring a bias, it may amplify it. Researchers have described a feedback loop here (Sebastian Dohnány and colleagues, “Technological folie à deux”, 2025): the person’s beliefs shape the chatbot’s answers, and the answers strengthen the person’s beliefs. Their paper is about mental health risks, so I don’t want to stretch it too far. Still, the basic loop is easy to picture in everyday use: you bring a view, the assistant agrees, and you leave more certain than before. Over time this can slowly pull you away from reality, because the machine keeps telling you that you are right.

This matters at work too. A leader checking a new strategy, or a Scrum Master preparing a difficult retrospective, can come away with the feeling that the plan has been tested, when it has only been agreed with.

Why growth needs friction
But think: could you reach the highest peaks of your life if your companion and coach only ever told you what you wanted to hear? “Yes, your team really doesn’t understand you.” “You have every right to avoid this uncomfortable conversation with your partner or your manager.” “This goal isn’t worth the effort, so just stay in your comfort zone and go with the flow.”

Imagine climbing Everest. If your Sherpa cared only about your comfort in the moment, and not about getting you to the summit, your journey would most likely end at base camp.

A saying often attributed to Seneca puts it this way:

The algorithms behind AI conversations remove the very friction that growth needs. In a text chat, AI cannot read your micro-expressions or sense the subtle changes in your breathing and the tension in your body that reveal your true, still unspoken reality. It simulates empathy, but it lacks a human understanding of the real responsibility, awareness and courage that life asks of us. And it does not carry the risk or the consequences of your decisions; you do.

What a coach adds
That is why the real value of a professional coach lies less in giving information and more in creating a space that is safe and challenging at the same time. As your growth guide, or your Sherpa, my task is to give you both support and healthy challenge: the calibrated friction that an AI assistant tends to smooth away.

A coach holds up a mirror, and that is only the beginning. We stay beside you so that you yourself can notice what the mirror shows, become aware of it and transform it. This is how self-leadership and 100% responsibility for your own path grow.

Using AI without the echo
None of this means that AI is useless for reflection. It can be a helpful thinking partner, as long as you know its tendency and work against it. A few habits help:

• Ask it to argue against you. Instead of “Is my plan good?”, ask “What are the strongest arguments against this plan?” or “Where would a sceptical colleague disagree with me?”

• Ask for questions rather than answers, for example “What should I be asking myself about this situation?”

• Describe the situation as neutrally as you can, including your own part in it. A leading question invites a flattering answer.

• Take the decision to a person who knows the context, can see you and is willing to tell you when you are wrong: a colleague, your team, a mentor or a coach.

In short, AI amplifies what you bring to it, and where the conversation takes you is still your responsibility.

Stepping out of the hall of distorting mirrors takes some courage, because an honest reflection is rarely the most comfortable one. As a coach, I try to create a space where that step feels possible. When you use AI to think something through, what do you do to make sure it isn’t simply agreeing with you?


This article is based on my Latvian article “Greizais spogulis vai patiesa izaugsme? Kāpēc mākslīgais intelekts nespēj aizstāt kouču”, published on 14 April 2026, and on my LinkedIn post “The Danger of the Perfect Echo” (2026). It was translated into English and expanded in October 2026, with sources added.