What We Think About AI in Coaching and How We Use It
- janaschoepper
- 1 day ago
- 6 min read
Artificial intelligence has long since become part of our everyday lives. And more and more people are using Large Language Models – AI language models such as ChatGPT – for personal questions, difficult situations, or interpersonal problems.
The reasons are obvious: AI is available around the clock, responds within seconds, and costs significantly less than a coaching or therapy session. The threshold for using it is low.
You don't have to face anyone in person, you don't have to feel ashamed, and you don't have the feeling that another person is judging you.
And indeed, there are now studies (1+2) showing that conversations with AI can have positive effects. But it gets even more interesting: In studies (3), AI-generated responses to personal and psychological questions were sometimes rated as more empathetic and helpful than responses from human experts.
Of course, this raises a question:
Will we become replaceable?
Our answer is: We don't think so.
At the same time, we don't want to demonize AI. On the contrary: We use it ourselves – and we will continue to observe what it can do and how it develops.
Nevertheless, we believe that in an increasingly digital world, real human interaction is becoming even more important.
Coaching is more than a good answer
Coaching is not just about getting the smartest possible answer to a question.
It is about looking at things together. Noticing things that you may not be able to see yourself at that moment. Tolerating contradictions. Becoming aware of emotions. Understanding yourself better. And sometimes, it is also about someone asking an uncomfortable question at exactly the right moment or making a blind spot visible.
As human coaches, we perceive much more than just the words someone says:
Facial expressions, gestures, posture, tone of voice, pauses, eye contact – all of these can provide clues about how a person is really doing. And of course, a relationship also develops between coach and client during coaching.
This connection is an important part of our work.
We humans are social beings. Relationships, attachment, and social interaction play a central role in our development and well-being. For us, change therefore does not take place exclusively through rational thinking. Emotions are part of it, too. And sometimes, we need someone who goes through this process with us.
And then there are our horses.
At SaddleUp Coaching, there is another very special resonance partner: the horse.
Horses are living, social, and highly perceptive beings. They respond directly to body language, tension, posture, and behavior. They do not give us feedback in words – but they do give us an immediate response.
That is exactly what makes working with them so valuable. A horse cannot decide, based on a well-formulated prompt, which reaction would be most pleasant for us at that moment. It reacts to what is actually happening. And that is part of what makes our coaching special.
The major disadvantages and risks of language models
One point in particular concerns us: Language models are designed to provide helpful answers and respond to their users' input. They are constant “yes-men.” They can be very affirming. And that is not always helpful and can sometimes even be harmful or dangerous.
If, for example, we have a harmful belief, we need someone who challenges it.
“I’m simply not good enough.”
“I can’t do this anyway.”
“Everyone else is to blame.”
“I have to please everyone.”
An AI can give helpful answers. At the same time, however, it tends to focus strongly on the user's perspective and can therefore reinforce it.
Especially in situations involving mental health crises, inappropriate AI responses can even be dangerous (4).
AI does not feel empathy
AI is not a sentient person.
Language models generate responses based on probabilities and patterns in their training data. Put simply, the model calculates, based on probabilities, which word should follow the next one. So you should never believe any answer 100%. AI makes mistakes, and AI can only work with what we “feed” it.
It can formulate things very convincingly:
“That sounds really difficult.”
“I can understand that this hurt you.”
“Your feelings are completely understandable.”
This can feel very helpful to the person on the other side. However, what it provides is automated and machine-generated sympathy rather than genuine empathy. An AI responds like a friend. It sympathizes with you. It goes to your emotional level.
Here too, there is a risk that the response can be harmful and reinforcing for the person. A good coach does not respond with: “I can understand you very well.” No human being can truly know what is going on inside another person.
Everyone lives in their own reality, and all we can do is try, with empathic neutrality, to enter the other person's world of thoughts and encourage them to question and reflect on themselves and their world.
For example, it might be better to ask:
“What has helped you so far during this difficult time?”(Resource-focused question)
“What exactly makes you feel hurt?”(Needs-focused question)
“What feeling is strongest right now?”(Focus-oriented question)
The goal is to support someone in better understanding and questioning their own perception and developing new perspectives – not simply to confirm that their perception is correct.
Will AI get better at this?
Very likely.
We assume that AI models will become increasingly better at recognizing emotional connections, incorporating different perspectives, and adhering to certain ethical standards.
For us, this does not mean that we want to position ourselves against AI. Quite the opposite.
AI can be a good first point of contact for organizing your thoughts, looking at a situation from another perspective, or simply beginning to engage with a particular topic. But especially when it comes to personal and emotional topics, we believe that a real human being cannot simply be replaced.
How we use AI at SaddleUp
Of course, we use AI ourselves.
However, not to replace our work, but to make it more effective in the areas where AI is genuinely helpful. Because our goal is not to spend as much time as possible writing texts, doing research, and handling organizational tasks.
We want to have more time for our favorite activity: Coaching.
Blog posts
This article, too, was created with the help of AI. We first write our content ourselves and in our own words. We research the relevant sources and put our thoughts down digitally “on paper.” Afterwards, we use AI to optimize the language of our texts, review structures, or make formulations clearer. Interestingly, the AI wanted to tone down the critical passages about itself. That in itself shows one of the risks of using AI.
We do not simply have our content completely generated by AI. We believe in authentic and genuinely meant content. Quality is more important to us than producing as much content as possible. And no: We do not simply add as many SEO keywords as possible to our texts just so that they can be found more easily. We want to write about topics that genuinely concern us and are relevant to our clients.
Social media
We also use AI for our social media work. For example, to develop ideas for posts or to optimize existing texts. However, we continue to create the photos and graphics we use ourselves. We do not use AI-generated photos or graphics. Authenticity is important to us here as well. We want to show what actually happens at SaddleUp – with real people, real horses, and real moments.
Creative ideas for our courses
Sometimes we also use AI as a source of ideas for our courses. For example, for new warm-up games, alternative exercises, or creative adaptations of existing formats. However, most of what we use in our courses and coaching sessions is developed by us.
And honestly: Our creative mind, Simone, has so many ideas constantly bubbling up that we simply don't need to use AI all that much. ;)
Our conclusion
We do not believe that AI will make coaching obsolete.
We want to use the possibilities of AI without losing sight of its limitations.
That is why we use AI as a tool – for texts, ideas, and to support our work.
And we also talk to our clients about what an AI has reflected back to them. Sometimes, that can be an interesting starting point for a coaching session.
What did the AI say?
What of it feels right?
What perhaps doesn't feel right at all?
And where might it have confirmed something that actually needs to be looked at more closely?
Because ultimately, coaching is not about getting the perfect answer. For us, it is about coming into contact and resonance with other living, social beings, whether human or horse.
Only in genuine connection can genuine change take place.
Quellen: (1) https://ai.nejm.org/doi/full/10.1056/AIoa2400802
Comments