Stellenbosch Business School research finds that AI technology rivals human coaching Stellenbosch Business School Skip to main content
AI technology
The benefits of coaching, helping individuals and organisations to achieve their goals, are out of reach for many, due to high costs.

The benefits of coaching, helping individuals and organisations to achieve their goals, are out of reach for many, due to high costs and scarcity of skilled coaches – but new research showing the effectiveness of an Artificial Intelligence (AI) coach called Vici is set to change that and democratise the business of coaching.

A 10-month study at the University of Stellenbosch Business School (USB) found that the chatbot Vici, available free for download, which deploys evidence-backed coaching strategies to interact with users, helped people to achieve their goals at double the rate of a control group attempting to reach their goals on their own.[i]

And the “pleasantly surprising” finding of a follow-up study was that Vici was just as effective as human coaches,[ii] said study leader Dr Nicky Terblanche, head of the MPhil in Management Coaching programme at Stellenbosch Business School.

Dr Terblanche’s research findings represent a ground-breaking study in the effectiveness of AI in coaching as they are the first to track the development and testing of an AI chatbot-coach that is scientifically designed based on research findings on what makes human coaching most effective.

The chatbot is called “Vici” after the Latin phrase “Veni, Vidi, Vici”, commonly attributed to Julius Caesar, meaning “I came, I saw, I conquered”.

Vici (www.coachvici.com) helps users to identify goals, break them down into short- and long-term goals, specify the actions they will take to reach their goals, monitor their progress and adjust goals or actions where needed. The AI coach is accessed via the Telegram messaging app or Facebook Messenger and, unlike human coaches, is available 24/7.

Dr Terblanche said the group using Vici had a 55% increase in goal attainment, compared to a 24% increase in the control group who were attempting to achieve their goals on their own without the help of the chatbot. The more frequently the users interacted with Vici, the more they increased their goal attainment, and three months after ending their chats with Vici, they still reported a continued momentum in achieving their goals.

The research into Vici’s effectiveness replicated a previous randomised control trial (RCT) study on the efficacy of human coaches, which found that human coaches were able to improve their clients’ goal attainment, psychological wellbeing and resilience, and reduce their stress.

Dr Terblanche said when the results of human coaches vs Vici were compared, human coaches outperformed Vici on increasing psychological wellbeing and resilience, and lowering stress, but on goal attainment – the single function which Vici was designed to perform – the chatbot performed just as well as the human coaches.

He said the results were a signal that, while AI is not yet advanced enough to replace human coaches entirely, it could effectively replace those using simplistic, narrow approaches, and that AI coaching could be “scaled to democratise coaching and provide its benefits to people in regions of the world where coaching services are scarce and expensive”.

Organisational coaching – which encompasses genres including executive, workplace, managerial, leadership and business coaching – has been proved through research as “a mechanism to support people’s learning, growth, wellness, self-awareness, career management and behavioural change”.[iii]

One of the fastest growing “helping professions” in the world, coaching contributes positively to employee and organisational performance,[iv] but a lack of skilled, qualified coaches and the high costs due to scarcity put the service out of reach of the majority, especially in developing countries.

Dr Terblanche said statistics from the largest professional coaching body, the International Coach Federation, showed that fewer than 2 000 of their 42 786 members (less than 5%) in 2020 were practicing in Africa. Meanwhile, the average cost of executive coaching in Africa is around R1 700 an hour, and R1 500/hour in South Africa.[v]

“This combination of low availability and high costs places human coaching well beyond the reach of most people in Africa.

“AI coaching has the potential to deliver personalised coaching at scale, reduce costs and reach a wider audience across various domains, with 24/7 availability. Now with its efficacy proven in our study showing that the AI chatbot coach Vici was able to successfully assist users in increasing their goal attainment, there is a compelling case to employ AI coaching to democratise this helping service and extend it to previously excluded societies,” he said.

Conventional coaching is usually a one-on-one structured conversation between coach and client, although Dr Terblanche said that even before the Covid-19 pandemic and the rise of remote meetings, only about a third of coaching interactions were taking place face-to-face in a physical space. [vi]

“The use of technology in coaching is not new, with human coaches often using digital platforms such as WhatsApp, FaceTime, Skype or Zoom to interact with clients instead of face-to-face sessions. Digital coaching platforms already exist, matching employees with coaches, hosting online coaching sessions, and providing data analytics to the employer to track progress.

“But all of this still relies on human coaches, which come at a cost and with challenges of availability and accessibility.

“We sought to provide a solution that would be affordable, available and accessible – and evidence-based, backed by research,” Dr Terblanche said.

He said Vici was not intended to replace human coaches and face-to-face interaction.

“To match a human coach would require a Strong AI entity but this field of research is in its infancy and it is highly unlikely that we will see an AI entity able to convincingly perform the functions of a human coach any time soon.”

He said the Vici chatbot is an example of “Weak AI” or Artificial Narrow Intelligence (ANI), in that it is focused on a specific narrow task and not a wide, general intelligence with ability to learn. It is a type of “expert system” programmed with specialised knowledge and “able to provide acceptable solutions to individual problems in a narrow topic area”.[vii]

While coaching has numerous proven benefits including improved psychological wellbeing, reduced stress, better work relationships and management, Vici was designed to fulfil one specific coaching task, namely goal attainment, because supporting achievement of goals is a hallmark of coaching and what sets it apart from other helping professions.[viii]

Although little to no research has been done on the effectiveness of AI in organisational coaching, Dr Terblanche said research in related fields such as psychotherapy and healthcare had shown the way. Previous research had shown the effectiveness of chatbots in supporting behavioural change, helping people to set and achieve goals for self-improvement, for example in areas such as eating habits, depression or getting more exercise.[ix]

Dr Terblanche, who started his career in software engineering, specialising in artificial intelligence, before moving to lecturing and research in coaching at the Stellenbosch Business School, said this combination of interests and a passion for putting research and technology in the service of helping people put him on the path to developing Vici.

He first developed the Designing AI Coach (DAIC) framework to guide the design of artificial intelligence-driven coaching applications based on scientifically proven coaching models and methods, international standards of coaching practice and ethics, and best practices in AI and chatbot design.

Working with a team of professional coaches and computer scientists, he then used the framework to develop Vici and tested its effectiveness in a 10-month randomised control trial (RCT) study, with 268 participants – half receiving AI coaching from Vici via the Telegram messaging app, and the other half a control group receiving no coaching.

Available 24/7, Vici helped users to identify goals, specify the actions they would take to reach their goals, monitor their progress and adjust goals or actions where needed.

The experimental group using Vici had a 55% increase in goal attainment, compared to a 24% increase in the control group who were attempting to achieve their goals on their own without the help of the chatbot. The more frequently the users interacted with Vici, the more they increased their goal attainment, and three months after ending their chats with Vici, they still reported a continued increase, while those who didn’t use Vici, reported a decline in their goal attainment in the three months after the experiment had ended.

Dr Terblanche said the monitoring aspect of Vici – with all interactions after the initial goal-setting conversation being focused on monitoring progress – explained how the “relatively basic” chatbot had such a significant effect on the users attaining their goals.

“Monitoring positively influences progress towards goals by helping to translate them into actions and then tracking progress. The frequency of progress monitoring also has a positive effect on goal attainment and the results of this study, where participants who interacted with Vici more frequently made more progress towards their goals, shows that this is also true for AI coaching.

“Studies on human-to-human coaching have also shown that physically recording goals, rather than just making a mental note, increases goal attainment and this proved to be true with Vici as well, where users had to physically type their goals into the chatbot system, creating a written record and commitment.”

Goal theory also helps to explain Vici’s effectiveness and impact, Dr Terblanche said.

“In the goal-setting and progress conversations, Vici asks explicit questions that help users to continuously assess how realistic their goals are, how committed they are, and if they have the resources they need. Vici also help users to break their goals down into long-term and short-term, because progress and feedback on short-term goals helps keep the person realistic and on track towards the long-term goal.”

Dr Terblanche said the effectiveness of Vici had validated the approach of designing the chatbot in line with the DAIC framework that is based on sound evidence, research and best practice AI design.

Explore AI coaching with Vici the AI Coach at https://coachvici.com/, with links to connect with the chatbot on Facebook Messenger or Telegram. Sessions are free of charge and available 24/7. All data

The website also gives access to an international panel of professional human coaches available for paid online coaching sessions.

[i] Terblanche, N. 2021 with Dr Joanne Molyn, Prof Erik de Haan and Viktor Nilsson. Artificial Intelligence coaching, it works! The impact of AI coaching on goal attainment. Stellenbosch Business School study. Pre-print study.

 

 

[ii] Terblanche, N. 2021 with Dr Joanne Molyn, Prof Erik de Haan and Viktor Nilsson. It finally happened: An Artificial Intelligence coach rivals human coaches. Pre-print study.

 

[iii] Terblanche, N. 2020. A design framework to create Artificial Intelligence Coaches. International Journal of Evidence Based Coaching and Mentoring. https://radar.brookes.ac.uk/radar/items/312d40ec-ccdf-431c-a062-2aa862166ac4/1/ or direct link to article in PDF at www.doi.org/10.24384/b7gs-3h05

[iv] Reference 1.

 

[v] Terblanche, N et al. 2021. African organisational coaching practice: Exploring approaches used, and the factors influencing coaches’ fees. South African Journal of Business Management. https://doi.org/10.4102/sajbm.v52i1.2395

 

[vi] Terblanche, N & Cilliers, D. 2020. Factors that influence users’ adoption of being coached by an Artificial Intelligence Coach. Philosophy of Coaching: An International Journal. https://philosophyofcoaching.org/ (Click on Volume 5, No 1, May 2020), or direct link to article PDF at http://dx.doi.org/10.22316/poc/05.1.05

[vii] Reference 1.

 

[viii] Reference 1.

 

[ix] Reference 3.

Subscribe

Want to stay in touch with the Stellenbosch Business School community? Sign up and receive newsletters from our desk to your inbox.

SIGN UP