AI and coaching – opportunities and challenges

This January I met Vicci, a coachbot, at a conference. When I read about Vicci in advance, I assumed I would be meeting a human-like coaching robot. Yet at the conference, Dr NickyTerblanche from the University of Stellenbosch Business School in South Africa demonstrated a computer programme that mimics human conversation via a text interface, called Vicci.

One of the success factors for the adoption of chatbot technology is clarity that the bot is not a human so as to manage expectations. Vicci is underpinned by a theoretical model derived from goal attainment and performance coaching theory. In a conversational style, Vicci helps clients to identify goals, reflect on their importance and then create an action plan to reach the goals. Regular check-in sessions track the progress of the actions and goals.

Vicci is ready for research purposes and Dr Terblanche plans to conduct a number of investigations ranging from willingness of clients to use Vicci (technology adoption), to the efficacy of Vicci in terms of helping clients reach their goals.

Benefit of using AI in coaching

Due to the self-directed nature of many AI-based coaching offerings that are already on the market, it makes coaching more accessible to a wider group of employees. Younger generations joining the workforce value continuous performance feedback and many organisations have moved away from annual performance reviews to ongoing feedback.

Some AI coaching services are integrated with feedback tools and provide a combined service of coaching based on feedback collected. AI and technology are changing increasingly how organisations are managed. For instance, executives will have to manage a hybrid workforce made up of humans and AI-based services. This new reality can be
mirrored in coaching based on or complemented by AI to provide an additional learning opportunity for developing relevant skills and experiences as a user.

While humanoid coaching robots exist as well as text based chatbots, a binary discussion on whether AI-based coaching will replace the human coach is not necessarily helpful, as there is much potential for complementary use. David Clutterbuck sees AI providing information and suggestions to the coach as the key to the successful use of AI during coaching sessions.

These include emerging linguistic patterns, useful strategic planning models and decision trees for complex choices. (AI can also identify signs of sociopathy in conversations!). However, this three-way approach needs to be managed: the coach might be distracted by the information flow and the client may feel left out if there is a strong partnership between the AI support and coach. Therefor Clutterbuck calls for a three-way partnership in which the client is also able to access AI.

AI could furthermore be used in-between coaching sessions, e.g. Vicci could complement the advantages of the human coach-client interaction by focussing on the goals aspect. Evidence of the successful use of a ‘bot’ to complement human interaction is available from the field of mental health: the ‘Woebot’ is an automated chatbot, based on cognitive-behavioural therapy (CBT), for young adults with symptoms of depression and anxiety. The free-of-charge Woebot is positioned to the users as ‘coaching’ and ‘self-help’.

The debate about differences and overlap of coaching and mentoring is ongoing. AI works on so called ‘training data’ it has been taught, which aligns more with the importance of past experience in mentoring. Thus, there is a lot of potential for using coachbots in mentoring contexts as well as in coaching. How coaching is used is diversifying rapidly in organisations generally which therefore calls also for a diverse offering of AI-based coaching. Many formerly advice-giving services are now adopting a coaching approach, e.g. health and career guidance.

Organisations frequently offer 1:1 or group coaching in support of career and well-being goals and coachbots are available that can complement this. Further, some AI-based skills coachbots are used integrated with other learning interventions, e.g. communication coachbots as part of leadership programmes.

For professionals procuring coaching, AI coachbots require assessing new kinds of services and providers. It would be interesting to test how coachbots perform against the coaching competency frameworks used for accrediting coaches! Coachbots are usually available through subscription services, but organisations can also build their own ‘bots’ with the help of experts.

Challenges

Dr Terblanche’s view on AI is that currently the ability and impact of AI is hyped beyond what it can deliver. Instances such as Vicci represent artificial narrow intelligence (ANI) where a computer can perform tasks requiring a rather rudimentary form of intelligence compared to humans. The hype around AI is often based on artificial general intelligence (AGI) where a computer is equal and superior to human intelligence. We are still very far from anything close to AGI and some commentators say that we will not see true AGI in our lifetimes.

So, what does the future hold for AI and coaching? “Coaches and HR practitioners should educate themselves on the current state and ability to AI in their field to be able to distinguish between hype and reality. They should also become actively involved to shape the development of AI in their field to ensure adherence to ethical conduct by these entities,”
Dr Terblance advocates.

Issues with bias, such as gender bias in AI-based recruitment, have been widely reported, due to basing data on successful past applicants. Maximilian Hofer, PhD candidate in Machine Learning, explains: “At their core, AI-enabled applications make predictions based on historic patterns. We need to identify potential inherent biases and compensate them with a healthy dose of (human) judgement. Prediction and subsequent action are two separate tasks, which are connected by judgement. Judgement is a strength of humans and a weakness of machines.”

This means in the future employee training should shift from prediction-related skills to judgement-related ones. A key task for future managers is to build the most effective teams of judgement-focussed humans and prediction-focused AI
agents. Hence, when considering AI and coaching, it is not just the use of AI within coaching itself, but also the increased use of AI in the wider organisational context of the coachee. How will this change how coaching is used generally in organisation and what topics arise in executive coaching?

Returning to the issue of bias, it is more complex than just being caused by biased data. Bias can creep in at many stages of the machine learning process underlying many of AI’s applications. Starting even before data is collected, there is the decision-making on what the deep-learning models should achieve. In order to manage bias and ethics, Dr Paula Boddington, Research Associate at Oxford University calls for as many people as possible with diverse experiences and viewpoints to be involved in AI. For example, Dr Sandra Waechter, a lawyer and Research Fellow at the Oxford Internet Institute, points out that: “Law is a very male-dominated field and tech-law even more so. The general view of what a tech-lawyer ‘is’ is not very diverse or evolved yet.”

In summary, AI will change the organisational context for all coaching clients. There are many advantages and opportunities where AI can complement coaching. Self-directed access through Apps makes AI-based coaching more accessible to wider employee groups. There are many challenges associated with AI-based coaching around managing bias and ethics and we are called to contribute to managing these challenges by educating ourselves, contributing to research and ensuring diverse viewpoints are represented in the field of AI-based coaching, but also as AI is evolving generally.

*This article was first posted on the Executive Coaching Consultancy’s website.

• Vote  
• 1
• 0
• You already voted!