Ethics in digital and AI coaching

ABSTRACT

Coaching in a digital environment has become increasingly popular, especially during the pandemic, with several potentials in terms of accessibility, adaptability, and flexibility. In addition, the first artificial intelligence (AI) tools have emerged as coaching interventions for self-development and well-being. However, the complex nature of the digital environment coupled with the unregulated nature of the coaching profession pose a significant risk for ethical mismanagement. This paper discusses the potential ethical risks in digital and AI coaching that can arise from the (non)human coach, the client, a third party, or the (digital) coaching context. Derived from this, the need for coach training and supervision as well as a sound coach selection based on ethical competencies is emphasized. In addition, the limitations of AI coaching are addressed, suggesting an integrated approach of AI alongside a human coach.

KEYWORDS:

• Coaching
• ethics
• digital coaching
• AI coaching

Introduction

Coaching has gained significant popularity as a human resource development (HRD) intervention over the past two decades, with an increasing emphasis on digital coaching, particularly since the COVID-19 pandemic (International Coaching Federation (ICF) 2021a; Passmore et al. 2023). Thus, the digital coaching market increased in recent years with software providers, platforms, and even AI coaching offers (Passmore and Tee 2023). This increase in interest can be explained by several advantages digital coaching can offer for the coaches and clients that among others tackle cost and time efficiency, accessibility, as well as independence from geographical distance (Kanatouri 2020; Passmore et al. 2023). Furthermore, AI can offer several benefits for clients, such as the independence of a human interaction partner, enhancing the aspect of cost efficiency and flexibility (Allemand and Flückiger 2022; Allemand et al. 2020; Passmore et al. 2023). Digital and AI coaching both explain a coaching approach using digital technologies for learning and development that is either with a human coach (digital coaching) or a non-human machine-assisted process (AI coaching) (Diller and Passmore 2023; Graßmann and Schermuly 2021). This digital / AI coaching approach is different to other digital / AI HRD interventions, such as training or consulting, as coaching is a more individualized approach with the coach-client relationship and the client’s self-reflection and self-development at its core (Diller and Passmore 2023; Diller et al. 2021).¹

While digital and AI coaching may come with several benefits, they also pose great ethical risks, regarding, for instance, inter alia data security and confidentiality (Kanatouri 2020; Passmore and Tee 2023). The present paper explores the benefits and ethical risks of digital and AI coaching as well as presents practical implications to meet ethical standards in the digital environment. For this, a conceptual literature review was undertaken based on the unique nature of the research area, which is marked by a scarcity of pre-existing literature due to its relative novelty. A conceptual literature review is an examination and synthesis of existing literature that focuses on the theoretical or conceptual frameworks within a particular field of study. Unlike a traditional literature review, which may primarily summarize and analyze empirical studies, a conceptual literature review delves into the theoretical foundations, key concepts, models, and frameworks that underpin research in a specific area (Callahan 2010; Fan et al. 2022). This approach was chosen to not only include literature on digital and AI coaching but to further transfer established knowledge from digital HRD, ethics in the digital world, ethics in overall coaching, and other areas to attain a comprehensive understanding and transfer vital insights and data crucial for a thorough exploration of this evolving field.

The benefits of digital and AI coaching

Coaching as an HRD intervention has a widespread appeal that is underscored by meta-analyses and meta-reviews in support of its beneficial effects, showing not only effects on an individual but also on a team and an organizational level (DeHaan and Nilsson 2023; Grover and Furnham 2016). In addition, first exploratory studies show promising effects of digital coaching: Digital coaching and coaching-related digital interventions seem to positively influence measures of coaching success, such as well-being, goal attainment, and coaching satisfaction (Bak et al. 2023; Carson, Choppin, and Choppin 2021; Ghods 2009; Kim and Lee 2023; Poepsel 2011; Wang, Dai, and Chao 2023). Furthermore, a first comparison study between face-to-face and distance coaching points out that the working alliance as an essential basis for coaching success can be established in a digital coaching environment (Berry et al. 2011). In addition, coaching-related AI interventions seem to benefit the client’s goal attainment and self-control (Allemand and Flückiger 2022; Nicky et al. 2022).

On top of these preliminary findings, the digital environment can offer additional benefits for both the client and the coach. Digital coaching is easily accessible, flexible in time and place, and, thus, can be cost-effective in terms of both time and finances (Brandenburg and Ellinger 2003; Frazee 2008; Hernez-Broome et al. 2007; Kim et al. 2005). For instance, clients and coaches do not need to book and travel to a coaching room. Furthermore, digital technologies enable the monitoring of thoughts, feelings, behaviours, and processes, which can provide valuable insights for clients and can facilitate coach supervision (Allemand et al. 2020; Amichai-Hamburger et al. 2014; Harari et al. 2020). In addition, digital environments can be perceived as safer and more secure spaces than face-to-face environments, which encourage open communication about sensitive issues (Domna, Hanumanthu, and Slater 2016; Miyahira et al. 2012). Digital environments further offer new opportunities for self-development, such as using avatars to enhance self-perception and explore different roles and identities (Mel et al. 2010; Normand et al. 2011; Yee and Bailenson 2007). In addition to the digital environment, AI coaching can not only further increase this accessibility, flexibility, cost-effectiveness, and feeling of anonymity: It can also adapt as a coach character to the client`s needs as well as can learn from big data on how to best react in situations (Graßmann and Schermuly 2021). These benefits highlight the potential of digital technologies to enhance coaching accessibility, experiences, and outcomes.

The ethical risks of digital and AI coaching

Nevertheless, there are various aspects of concern associated with digital and AI coaching (Kanatouri 2020; Passmore and Tee 2023). For example, coaches described the digital context as more challenging and more complex, as well as needing more engagement compared to other digital HRD contexts. Furthermore, coaches raise concerns about the loss of nonverbal communication and difficulties in establishing a secure and trustful relationship, as well as the risk of technology to disrupt, distract, or limit certain resources (Diller and Passmore 2023). These difficulties experienced by coaches offering digital coaching services give a first glimpse into how ethical issues might arise in the digital environment.

Overall, ethics in coaching has been a main subject of concern due to its harmful consequences for clients, coaches, and organisations (Brennan and Leni 2010; Hannafey and Vitulano 2013; Walsh Richard 2015). Various ethical dilemmas can occur in coaching situations with most coaches not knowing on how to deal with them (Biquet 2020). Such ethical dilemmas surface when neither of the available options seem to align with ethical principles (Turner and Passmore 2018). Besides dilemma situations, overall ethical issues can arise that can trouble coaches even when there is a clear ethical option (Diller et al. 2021).

A recent qualitative study with 264 coaches worldwide shows that such ethical issues including ethical dilemmas can arise from the clients, a third-party, the coaching context, and the coaches themselves (Diller, Passmore, and Frey 2023). The Model of Ethical Interdependence in Coaching (Figure 1) depicts these four factors and their dynamics.

Figure 1. A model of ethical interdependence in coaching (Diller, Passmore, and Frey 2023).

This model is based on the Loop-2-Loop Model of Social Interaction, which demonstrates how ethical principles of fairness play off in the dynamics of a social interaction (Jonas 2015). Thus, the model visualizes where ethical issues can arise and how they can develop within the dynamics of social interactions. As shown in Figure 1, ethical issues can emerge from the client`s behavior, such as the client behaving unethically by, for instance, making sexual advancements towards the coach. Additionally, ethical issues can result from the coach, such as having difficulties as a coach to stay in role due to disliking the client or liking the client too much. Furthermore, ethical issues can arise from a third party, such as a supervisor that wants to have a say on what the coaching should achieve with the client. Last but not least, ethical issues can stem from the coaching context, such as an unsafe or inaproppriate coaching room. These factors can influence each other in the social context of coaching. For example, an inappropriate third-party action may lead to an inapropriate client reaction. Furthermore, the factors involving human beings have their own inner worlds from needs and values to motivation that can influence their behavior in the social interactions. For instance, a person`s inaproppriate behavior may have had positive intentions. Reversely, if a person`s needs are threatened, they might react differently than they would in a non-threatening situation (Diller, Passmore, and Frey 2023). In sum, a variety of ethical issues can arise in coaching that can emerge from diverse interdependent factors.

However, most coaches do not receive specific training in applied ethics and further often believe in themselves to act ethically without possibly reflecting on this. Thus, the potential for ethical mismanagement is high with adverse outcomes for the overall coaching dynamic, ranging from distrust to severe legal consequences (Passmore and Lance 2011). This risk of ethical mismanagement is further increased by the unregulated nature of coaching: Anyone can call themselves ‘coach’ and offer coaching or even coach training, allowing anyone to assume the role without appropriate qualifications or training (Kühl 2021; Sherman and Freas 2004; Siegfried, Möller, and Wolfgang 2018).

These ethical challenges not only extend to the digital environment but can further introduce new ethical difficulties. For instance, communication in digital coaching can lead to ethical issues of transparency, disclosure, and privacy. This tackles issues of how notes are taken, where information is stored, and whether sessions are audio/video/chat recorded (Hawley, Eve, and Ioanna in press; Iordanou, Hawley, and Iordanou 2017). Furthermore, the digital space poses further risks of data security mismanagement. These risks occur concerning personal data from the client, the coaching session (e.g. recorded session), and personal data from the coach. In addition, digital files can carry viruses, computers can get hacked, and not every video-chat platform focuses on data security (Fouracres 2022, 2023). Therefore, the risk of information that should have stayed sealed being disclosed is higher compared to face-to-face coaching (Kanatouri 2020). In addition, managing emotions, interpersonal closeness, and a social connection in the digital coaching environment is itself a challenge (Meyer 2023). However, managing (strong) emotions or even ‘meltdowns’ as well as misunderstandings can become (ethically) challenging in a digital environment, as you do only have a limited space on how to be there for the client (Doyle and Sirivanasan 2023). Thus, emotions of sadness, anger, and distress in coaching can be ethically challenging to manage online. A high level of empathic accuracy is needed to first detect an emotional situation correctly as well as react to it appropriately, which can cost a lot of coach attention and energy (Hawley, Eve, and Ioanna in press). Responding appropriately to the emotional state of a client might be especially challenging for AI coach bots (Passmore et al. 2023). In addition, ways to intervene in an emotionally challenging situation are limited in a digital environment: If a client has a break-down on the other end of the line, a coach cannot make a tea, give a tissue, put a hand on a shoulder, or in other ways show their sympathy and empathy in person, leaving it up to showing empathy in mimicry and emotional perspective taking. This approach might work in a situation of deactivating affective states, such as sadness and depression, but might be difficult with activating affective states, such as anger, nervosity, and distress (see Circumplex Model; Jonathan, Russell, and Peterson 2008).

Moreover, digital coaching environments can be discriminative, as they are often designed for a specific group of healthy and technically skilled people. For example, people with neurodiversity may have difficulties in sitting still and looking at a screen and themselves without breaks; low-quality equipment, such as a bad microphone can enhance these sensory stress situations (Doyle and Sirivanasan 2023). This aspect of discrimination is of particular importance when it comes to AI coaching, as AI has been shown discriminatory behaviour in the past: For example, AI showed racial bias in health treatment neglection and job application rejections (Köchling et al. 2021; Obermeyer et al. 2019; Shi et al. 2020). Similarly, discrimination based on gender is prevailing in many AI systems, leading to favouring men (Leavy 2018; Singh et al. 2020; Smith and Ricanek 2020). These discriminations are based on both the data being mostly on white heterosexual men and the decision-maker being the same group (Nuseir et al. 2021; Perez 2019). In addition, working with AI proposes additional ethical risks, as making autonomous decisions beyond the guidelines provided by programmers can raise questions about accountability and ethics (Etzioni and Etzioni 2016). This issue is of particular importance when it comes to sensitive data and errors made by the system (Hermansyah et al. 2023). In sum, digital and AI coaching can hold several high risks of ethical mismanagement.

Practical implications I: ethical competencies as a must-have for coaches

To begin with, the ethical challenges faced in not only digital and AI but overall coaching highlight the importance of training and educating coaches in ethics (Maritz, Poggenpoel, and Myburg 2011; Passmore and Tracy 2020). Hereby, it is essential to move beyond normative codes that can be difficult or even variably to interpret to a more applied ethics approach (Brennan and Leni 2010; Diochon and Nizet 2015; Duffy and Passmore 2010; Iordanou, Hawley, and Iordanou 2017). For example, what does it mean to be ‘aware of and actively manage any power or status difference’ (ICF 2021b, 5) in terms of the behavior in a certain situation? Due to this abstraction, ethical codes can be unclear in situations (e.g. serving the client versus the stakeholder) or conflict with each other in other situations (e.g. confidentiality versus avoiding harm) (Passmore 2009). Thus, coach training and supervision is needed to gain applied ethical competencies.

Coach training and supervision

Coach training is not only more relevant for coaching success than experience or networks, it also enhances the coach’s self-awareness and reflective practice (Diller et al. 2020). Thus, training in ethics can both enhance ethical competencies and further increase the awareness of ethical issues. This awareness can improve problem-solving abilities (Andersson et al. 2022; Jones 2009). Diller et al. (2023) hereby give an outlook on what a coach training on ethics could incorporate based on the qualitative answers of the coaches. In addition, reflective and reflexive practices contribute to a heightened awareness of ethical concerns (Iordanou, Hawley, and Iordanou 2017), making supervision crucial for self-reflection on ethical issues as well as for responding appropriately and learning from such situations (Day et al. 2008; Ratlabala and Terblanche 2022). Hereby, coaches can benefit from group settings based on the value of information sharing, best practices, and open-mindedness among coaches to move beyond isolation (Brennan and Leni 2010).

Selecting coaches based on ethical competencies

Recent studies have shown that coach training and reflective practice are essential for coaching success, as guided self-reflection and feedback support the coaches’ self-development (Diller et al. 2020). Furthermore, ethical competencies can be trained and measured (Claus, Eichenberg, and Dietrich 2009). Selecting coaches based on their coach training, reflective practice, and ethical competencies is thereforecrucial. Ethical decision-making and interview questions on ethics can help to select coaches based on ethical competencies (Passmore and Lance 2011; Pohling and Anja 2017).

Practical implications II: additional competencies needed for digital coaching

Besides ethical competencies, additional competencies are needed when it comes to digital and AI coaching in order to avoid ethical pitfalls.

Social and self-management skills

According to Hawley et al. (in press), managing ethics in digital coaching includes the coach’s curiosity, creativity, kindness, role modelling, and reflexivity. Besides these relational competencies, self-management competencies are needed as a coach: Considering the prevalent issue of Zoom fatigue, with video conferences leading to emotional and physical exhaustion (Konrath and Delphine 2016; Shockley et al. 2021), coaches need to prioritize self-care. For example, mental health practitioners and coaches found sessions via video communications tools more draining and tiring than face-to-face sessions (Diller and Passmore 2023; Feijt et al. 2020). Practicing mindfulness can in this context help with self-regulation, stress management, and reducing negative emotional reactivity (Keng, Smoski, and Robins 2011; Shapiro et al. 2006; 2007).

Data management skills

Furthermore, coaches need a solid understanding of data security and management in the digital realm (Fouracres 2022, 2023). Despite the apparent safety of digital environments, there is an increasing prevalence of cybercrime that exploded similar to our digital data creation and storage (Andre 2023; Fleck 2022). This poses risks not only to a coach’s business (e.g. reputational damage, business interruption, legal costs) but also to clients as potential victims (e.g. exposed information and psychological consequences of fear, uncertainty, distrust). Safeguarding one’s practice, clients, and client data is paramount in this digital age (Fouracres 2023).

Practical implications III: being aware of the limitations of AI coaching

Whilst coaching-related AI interventions can have a high potential (Graßmann and Schermuly 2021) and approaches towards responsible AI are investigated, it does yet function on its own without ethical risks (Schiff et al. 2020). While responsible AI coaching is in its development, an integrated approach of AI alongside a human coach could be a ‘powerful combination’ (Kanatouri 2020, 151), with clear signals to indicate whether the coach or AI is guiding the conversation at a given moment (Buergi, Mona, and David 2023; Graßmann and Schermuly 2021). This transparency by clearly stating that the conversation is conducted by a machine as well as by providing full disclosure about the chatbot’s purpose and limitations are hereby crucial (Mai and Rutschmann 2023). Furthermore, there is a need for AI to evolve towards greater inclusivity. This can be achieved by actively involving individuals from diverse backgrounds in the design and testing of new AI applications (Passmore and Tee 2023). By doing so, it is ensured that AI tools are more culturally and socially sensitive, better serving a broader range of people.

Conclusion

Digital and AI coaching offers increase, making it essential to talk about ethical challenges, such as data security, bias, and managing emotionally charged situations. Particularly due to the unregulated nature of coaching, ethical mismanagement can occur, which is why coach training, supervision, and selection is needed based on ethical practice and ethical standards. Based on the development stage of responsible AI coaching, an integrated approach with a coach and AI is emphasized to secure ethical safety, which can further complement human empathy and intuition skills with innovative AI.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. When it comes to sports coaching research, this difference may not be vital but it is in HRD (Iordanou 2018).