INKOVEMA Podcast „Well through time“
#220 GddZ
Coaching and mediating with ChatPGT support
How can artificial intelligence support us in our counselling work?
How can we use them for our counselling work?
In conversation with Stefanie Düll
Stefanie Düllis a systemic coach with a Master's degree in Counselling and Coaching (M.A.) and in Digital Transformation (M.Sc.). In her work, she combines counselling with new technologies. This is an experiment in integrating the „Inner Team“ concept from the Hamburg School around Friedemann Schulz von Thun into counselling work in a supportive way.
Small series: New technologies for mediation and counselling
Well through time.
The podcast about mediation, conflict coaching and organisational consulting.
Contents
Chapter:
0:06 Welcome to the podcast
0:45 Introduction to ChatGPT in coaching
5:10 The experiment with ChatGPT
7:30 Integration of AI in coaching
14:28 First results of the application
18:38 The future of integrative coaching
23:55 Important information for coaches
Summary of content
In this episode of the podcast „Rund durch die Zeit“ we talk to Stefanie Düll about the Innovative integration of artificial intelligenceChatGPT, in particular, into the coaching process. The integrative approach is a central theme: How can technology-supported coaching enrich and improve the interaction with clients? Stefanie, who works both in the IT sector and as a systemic coach, shares her experiences from an experiment in which she introduced AI into her consulting work.
Stefanie explains that her motivation comes not only from curiosity about new technologies, but also from the desire to better master the challenges that coaches often have to overcome in their work. She thinks that artificial intelligence can not only act as an organisational tool, but could also provide deeper insights into coaching processes. She hopes that AI will help to generate new ideas and improve feedback loops in coaching.
We discuss the specific challenges of using AI in coaching situations. Stefanie emphasises that the coach is responsible for deciding which parts of the process are delegated to the AI and that careful selection of the model is crucial for success. The example of the inner team makes it clear how this structure served as the basis for the experiment. …
After successfully conducting the experiment, Stefanie is positively surprised by the quality of the results, but emphasises that this is not the end of the development. Her findings show that AI can act as a valuable tool to gain deeper insights during the coaching process, as long as the coach keeps the overview and guides the client through the process accordingly.
We learn that there are some clients who are enthusiastic about this experimental approach, while others remain sceptical. Stefanie respects these different perspectives and emphasises the importance of voluntariness and motivation in coaching. The discussion also touches on how AI can open up new ways of accessing coaching, especially for people who find it difficult to open up to a human coach.
In conclusion, Stefanie encourages us to be open to the possibilities that technology offers and to keep experimenting to find out how AI can be usefully applied in coaching. It's an exciting and forward-thinking topic that encourages coaches to question their traditional methods and potentially develop new approaches.
Complete transcription
[0:00]That you have to look at how I can also interact with a client
[0:06]
Welcome to the podcast
[0:03]continue the integrative approach, so to speak. Welcome to the podcast Rund durch die Zeit, the podcast about mediation, conflict coaching and organisational consulting, a podcast by INKOVEMA. I'm Sascha Weigel and I'd like to welcome you to a new episode. And today it's all about new technologies, new technologies that we can use in our counselling work, that we want to use, that you may have to use in the future if something like what my guest today has tested, tried out or perhaps even pioneered becomes established.
[0:45]
Introduction to ChatGPT in coaching
[0:41]I'd like to start by welcoming Mrs Stefanie Düll to the podcast studio. Hello, Mr Weigel. Hello Mrs Düll. This is supposed to be about a series of tests, an experiment that you carried out.
[0:55]By carrying out coaching work with the help of chat GPT or artificial intelligence. And I think there's a lot to be said about the wild fantasies that are now emerging about how you might have spoken and manipulated the chat or had it done, how you had the client speak to you on a short or long leash, so to speak. We'll take a closer look at that. But before we do that, who are you, Mrs Dö? And how did you come to do coaching with the help of ChatGPT? Exactly, a few words about me. So how I come to this topic, I think that's also who I am. I worked in the IT sector for a long time. My original training was in business informatics, which is why I have a passion for IT and new technologies, but I also work as a systemic coach. And that's why I wanted to take a closer look at this combination of the two subjects, so to speak. You are a trained, certified coach, you work in this field and you have now done this because of your passion for technology. Would you say that?
[2:11]Based on your experience as a coach, yes, but we also have coaching problems that we could solve with the help of technology. Was that more of a problem-solving approach, so to speak, or was it simply curiosity to unite these two worlds that you bring with you? I would say a large part of it was definitely curiosity and also this topic, a lot is influenced by AI in everyday life. Why should coaching be excluded in the long term? That's why I was also interested in how AI could establish itself in the coaching market in a profitable way. And of course, I think there are already a lot of studies on what problems AI can solve in the coaching context at the moment. But that wasn't my main focus, I would say. If we assume that the listeners are above all colleagues, i.e. those who work in a very practical advisory capacity, so to speak.
[3:16]I like to do this in coaching or mediation sessions or in other counselling settings. What would you say are the problems that exist, are being solved or are being left unresolved in this work today that can be better addressed with the help of this technology called artificial intelligence and everything that falls under it, rather than working on them or leaving them as they are at the moment? So what do you hope to achieve with artificial intelligence in your work? I think there are many, many areas where it can definitely be put to good use.
[3:57]On the one hand, I think every one of us coaches or consultants does it with a passion for people, but then of course has all this other organisational work to do. And that's where AI can be super helpful. When it comes to somehow generating new ideas for clients, somehow providing inspiration, somehow using a feedback loop, a kind of small supervision, I'll say that in inverted commas, you can use AI to your advantage.
[4:28]As things stand today, at least from my perspective. So these areas already exist in any case. There is also further research or further development in the direction of supervision in parallel to an actual conversation, i.e. really using live supervision and AI to analyse additional aspects and provide the coach or consultant with them. I believe there are still many areas in which this can be done. There will also be help in the future. I would like to start with your experiment, so to speak, because it has the particularity, if I may say so, that you can analyse it directly.
[5:10]
The experiment with ChatGPT
[5:08]in the counselling work. So not so much now. And the problems are probably even more pronounced in the area of how do I organise my small business? How do I manage my administration and the administrative workload? Technology has always helped with this, of course, and everyone has their computer, smartphone and digital calendar.
[5:30]So all these things where it helps and where it will certainly continue to be helpful for us in the future. But you have dared to experiment and have directly consulted this artificial intelligence in an advisory capacity. As a coach who has never done anything like this before, what do you need to know about how it is structured? So how can you visualise it? Have you just put your laptop next to it and let it talk? So, now please coach my client for me.
[6:04]How does that work? So how can you visualise this in practice? In my approach, I wanted to look again specifically at how I can use AI in an integrative way in order to actively work together with the client, so to speak. And I picked out a model for my study, so to speak, the inner team. I think almost everyone knows that when coaches and counsellors listen. I also interpreted it because of its complexity and then I thought about what part of the inner team there is that, from my point of view, could perhaps also be worked on accordingly with an AI. And for me it was clearly this initial development process, where the client is really about setting up the person, not going into the analysis and looking at where my inner conflicts are, but just who is there? And the part of the inner team process, so to speak, was worked out by the clients with JGPT.
[7:08]They then shared the course of the conversation with me and then we had another joint analysis session where we really looked at what I was taking away from the inner team for myself, but we had already prepared this part with JGPT. I mean one thing at a time.
[7:30]
Integration of AI in coaching
[7:25]Firstly, the decision that it is only about integrating artificial intelligence. So ChatGPT will be integrated into your counselling work. You are still in control of the coach. It's not about this chatbot getting started on its own, so to speak, and you just standing behind or next to it, but rather integrating it into your counselling work.
[7:52]You choose a model, and that seems to be a necessity. So you can't just assume, or you didn't assume, that it has read everything, it is well fed, this chatbot from OpenAI, it knows it all.
[8:09]Why do you need to decide on a model, so to speak? I have only chosen one specific model for the studies because I am also of the opinion that the coach is the superior part in this model selection, who wants to ensure the process, so to speak, who must also have the process knowledge, which model can be helpful now and then must also decide for the model itself, which parts I want to work on with ChatGPT and of course I must also give all this to the client so that he can take this into account in his input and his start of the conversation with ChatGPT, so to speak. If I now assume that ChatGBT knows everything and I let the client do everything, then I would expect ChatGBT to carry out the complete model, but the analysis is kept very superficial and the client ultimately takes nothing away from it. That's why this selection of which model I give the client in this initial prompt, i.e. this start of the conversation with ChatGBT, also includes all the information about the model that ChatGBT and the client need to know in order to be able to use it profitably. These were important points that coaches don't realise when they ask themselves how to use a machine like this or how to use a tool like this. In other words, the reduction to a tool, to a concept, to a model basis, makes it possible to work in depth in the first place.
[9:38]And then you say that you not only have to prepare the client in this case, but also the machine with your own prompts, technically speaking, with prompts. In other words, you start by working with the tool in the same way as you would with the client, perhaps in a parallel situation, by explaining the basics to them, giving them some basic information so that they can then set off to work independently. What are the key questions or guiding principles in this initial prompt? So what do you focus the tool on?
[10:12]Let's take the example of Schulz von Thun's Inner Team, as far as we know it, or those who don't know it yet can read about it from the audience. Exactly. So a very important point in this prompt is, what are the quality criteria that I would now use to determine whether a person has been developed, so to speak. So really this first, I want a statement, I want a message, I want an appearance, I want an emotional state associated with it. These four factors have to be fulfilled so that ChatGPT also knows that I have to ask for precisely these four criteria, so to speak. These were crucial aspects that were included in the prompt, so to speak. The coach's attitude was also clearly formulated once again: what is my approach, because this can vary from coach to coach and the client should ideally also be given a standardised picture in the coaching sessions outside of ChatGPT. That's why I put it in there again. Am I systemically orientated or goal-oriented or what is important to me in my work, so to speak? These were the factors that had to be fulfilled in this prompt. Again with the reference to the analysis that took place afterwards, at what point should the conversation end?
[11:28]I also included the fact that a summary of what was worked out was to be created at the end, so to speak, that technology simply listed the people again so that the client could get an overview and reflect again, is this now what I feel comfortable with, where I see myself again. In other words, all this time-consuming work in coaching, the individual personality traits that function as an inner team for each person. There are probably three to five personality traits, or what are they called personas in the concept?
[12:08]Exactly, well, I think I had given three to five times, as a guideline, and then just personas, inner personas. Personas. And this development of the client with the help of a knowledgeable dialogue partner, normally the coach up to now, that is outsourced in terms of time, that is the idea? Exactly. And you instruct this tool knowing full well that it can speak really well, it can reproduce a lot of material that reads like knowledge, but you instruct it as if it hasn't done that yet. They tell you what is important to you as a coach, they tell you the criteria, ask about them, don't come back without having an answer. And then they say to both of them, and now they do it until next week, have a chat, so to speak, dear clients, here you have the tool, it will ask you questions and then we'll go through the chat history together. Hey, you who are listening to this podcast, if you like it, why don't you press five stars and leave feedback so that others who haven't yet listened to the podcast or found it can do so. And now we continue with the episode in the podcast gut durch die Zeit.
[13:37]Exactly, so I think if you were to enter inner team in ChatGPT, it could also explain these aspects without any problems. I think my main concern here was to ensure once again that the criteria that I personally find important as a coach are also fulfilled, because they form the basis, because you don't know exactly which source ChatGPT is accessing. I didn't go through the chat history together with the client, but I actually had it sent to me before the session to read over it to see, okay, where do I want to follow up in the subsequent session if I noticed something and, above all, to follow up not with, ah, I noticed in the chat-GBT conversation that you somehow did this and that, but really rather openly asked again to make sure that a preconception is not already somehow retained.
[14:28]
First results of the application
[14:26]or something, but is approached openly again. And that means, so to speak, that your prompt, which you developed beforehand, turns the ChatGPT tool, which is accessible to everyone, into your tool. So your client talks to a very special configured AI because you say that this and this is important to me and therefore the conversation will also go in this direction and query these criteria.
[14:51]Exactly. The client goes home and won't have it on their OpenAI access. How does it work technically so that they can edit or prompt an AI that receives their configuration? With ChatGPT, you can start and share this conversation and then, as soon as someone else opens it, either in their own account or in an anonymous account without being logged in, so to speak, they can continue the conversation, which is then only shared with them. So it is only shared with me again when you actively decide to share the conversation again. That's actually pretty clever for the application topic you've chosen. Yes, then you basically just had to answer the next question that ChatGPT asked. What went through your mind when you knew for the first time that I was about to be sent a chat history and I could see and read for the first time how it works when the client has done it alone with this AI? I was looking forward to the result, but also a huge amount of nervousness, because at the time I didn't know how well it would really work, how high quality the people who were being worked on would be. Yes, very nervous, I would say, in any case.
[16:19]And then the eureka moment, so to speak, was when you saw the content. I had the impression, and we are referring here, or I am, to a large extent to the essay that you then wrote about it. I had the impression that you were positively surprised, that you really thought, wow, I wouldn't have guessed that. Yes, I was pleasantly surprised. But to be honest, I was still a bit sceptical even after these shared conversations, because I still noticed that I really did a qualitative analysis of the personas and realised that, okay, from my point of view, they are not yet at the level of quality that I would expect in my coaching.
[16:58]But then I was super positively surprised, especially after the analysis sessions, that everyone really did leave with such a profitable insight and that I didn't notice any difference after this analysis. But I was still a bit worried after the course of the dialogue. Is it really the same level of quality at the end of the overall coaching process that I can otherwise offer my clients? And the quality level is not so much just what you then have as written results, so to speak, but what the client has gained for themselves, what they take away, what they have learnt or experienced. And you say that there is no difference whether someone uses this tool continuously, in pairs, without technical support or with technical support in the form you have described.
[17:52]Exactly, so after this development process, where it's really just the inner team, I think there are differences in quality. The survey of the participants also showed that there was less satisfaction with the development process after this first part, which is why the integrative approach was used, because after this second part, where the analysis and the discussion took place, there was no difference in satisfaction or in the quality of the personas. When I look at the fact that this was a tool and a first experiment and that you didn't just work in the lab with a white coat, so to speak, developing concepts and theories.
[18:38]
The future of integrative coaching
[18:36]but also provide practical coaching. How do you envisage development continuing in this direction?
[18:46]Exactly, so that was an experiment, but since then I have also continued this integrative approach in my own coaching programme if the clients are keen. So there are clearly some who are very optimistic about it, because then they can feel their way around a topic themselves. There are also others who are very sceptical and don't want to do it. I leave that open, so to speak, but then of course I do it with all kinds of other models and each time I think about what would be very beneficial for me and where a coach would be necessary. And I also think that from my point of view, at least in this world that we are entering, we have to look at how we can continue the integrative approach in the interaction with a client.
[19:37]In the long term, I believe that there will be more and more things that can perhaps be replaced or organised differently. So I think we simply have to remain very open in the world and see what new technologies there are that can be integrated. And if I imagine that you are now considering whether the client wants this voluntarily or not, they must somehow be asked the question. It also seems to me that it could be an intervention in itself. What is this offer for you, so here, dear client, we can do it like this, we can do it like this. From your perspective as a coach, do you have the impression that it could be a useful intervention to let the client do this for a few hours at home with the help of this AI or this prompt or this model until the next session, whereas it would be more of a negative intervention for others, who would be more likely to be repelled and say, I don't want this at all. So if you look at it not only in terms of voluntariness, so to speak, but also in terms of the offer as an intervention, you can see that…
[20:48]The potential, i.e. that there could be clients for whom this would be a good coaching intervention and not only.
[20:56]We are Yes a bit, when I the still so say may, with the Question, think me, in the Industry, can the a better Coaching make or perhaps even replaced us that. So we must Yes so to speak a little in the Evidence stand, the harms not, but Helps us. But it could Yes also simple one really good, sensible Form of intervention be, the we so so far even not knew, next to ours Questions, ours Structures, the we build. I believe, it comes very on the Clients natural to, ultimately. The Heavy of the Topic, the Self-reflection, the the Client already itself on the Day brings and before all also the Motivation, itself with the Themes to deal with. So there notices one Yes also, gives it Clients, the are there perhaps already deeper direct on the road and more motivated itself again much more on it to stay and other, the make the not so and before all, when I in ChatGBT in the own Room Independent on it work, then comes it natural very, very strong on it to, like motivated is the Client really in the Moment itself on it to work and more above to think about, because ChatGBT provides one Question, the can I now short answer with one Set. Since can I but also five Phrases attribute or above think about it, is then me left. Since happens stop not these intensive, I question again to, I question again to, because me is it not deep enough.
[22:19]And in favour must the Client stop somehow also willing be, motivated be, Topic natural fit.
[22:26]On the one Page have we the Presentation also, in the Case now also, the Machine can not so good follow up, them can not so good motivate, that the Participants also their Experiences, their Memories, their Assessments reveal. On the other Page know we also, that it many People gives, the Go not to other People towards and pour her Interior from. But I says times, in the evening or at night with Pain, AI chatbots all the more more. At least when we the Companies, but also already official Try etc. Faith give, the talk not only with their Dogs and Cats very intimate, but also with their computers.
[23:04]From the Wait here could it Yes quite be, that there also again itself one Opportunity opened, other this Offer to make, first times with one chat or also with one voice-driven chat so to speak itself the Themes to approach and afterwards to decide, yes, I give the one human Expert yet. See You also like this, that there Opportunities arise and that we not only so think should, the Machine makes the not so good like we People. Nope, I believe even, there reach I whole new People, like You also said have, with many, those it also lighter falls, there ins Conversation to go, as direct somehow before Face one other Person somehow the Whole to be distributed. Since is simple one lower Hurdle with the one or other then there, where I one Coaching process at all perhaps first times Start can through so a Concept.
[23:55]
Important information for coaches
[23:55]What is still important? What should Coaches know, when them the Direction at least think want and anticipate want?
[24:06]So on most important from mine View is this Topic, the Coach remains responsible for the Overall process. The means, the Coach remains responsible in favour, the AI to explain, the Model to explain, the Coach here to take with you. Because he remains responsible in favour to decide, which Model is because meaningful, in which Kind and Wise and which Parts can I because at all outsource. Naturally also this whole Topic Data protection, there simple the Educational work to make. And the is not, I store something simple only from, but I carry furthermore the Responsibility.
[24:37]But I believe also whole Important is, that we simple open remain, us there also times trust, so a Experiment quasi to be carried out, at to look, like can it because function. To there simple ours Practice also for ours Clients natural ultimately better to design. Mrs Düll, whole sincere Thanks to, that You itself trusted have, this Experiment to be carried out. And many Thanks to also, that You it with us shared have here in the Podcast. Me is today some clearer become and also really more imaginable. Many Thanks to. The pleased me. Thank you also. Until to the next Paint to the second Experiment and one good Time for You. You also. Beautiful Day yet. Yes, the was my Conversation with Stephanie Dö to the Attempt, ChatGPT in the Coaching to integrate. And like them the made has, has them set out. I thank you me, that her with thereby maintained here with this Podcast and say goodbye me with best Greetings. Leave a message one Star rating, comes good through the Time. I am Sascha Weigel from INKOVEMA, the Institute for Conflict and Negotiation management in Leipzig and Partner for professional Mediation and Coaching training programmes.
Leave A Comment