How Will AI Change the Future of Coaching?

The arrival of OpenAI’s product, Chat-GPT, took the world by storm. Since then, a stream of new AI products have appeared on the market, from Google’s Bard, Meta’s LLaMA (Large Language Model Meta AI) and Elon Musk’s X.AI. Coaching providers, and some coaches, have rushed to operationalise generative pre-trained programs like GPT-4 to create Avatar coaches or conversational bots. After testing them all, it is clear that these products have great potential, but they are a long way from a high-quality coaching conversation.

Some of these bots have a tendency to give advice of varying quality, which is sometimes based on misinformation. Others, through good quality prompts, have had their advice preference curtailed, and on the whole, are able to ask open questions. But they still lack the skills of a human ICF (International Coaching Federation) or EMCC (European Mentoring and Coaching Council) certified professional coach (1).

For example, most bots are underdeveloped in their skills at summarising the conversation to provide a platform for their next question. They are less skilled at observing conversational patterns across the conversation (or multiple conversations) on which to base their feedback. Critically, they also lack empathy and instead offer responses which users experience more as ‘sympathetic’.

Of course, the may be different. In a year or two, we will likely have better bots, but the jury is still out as to whether managers will prefer talking to a bot over a human. One influencing factor will be pricing; if bots have a revenue model more akin to a Google search (free at the point of use), this might make them more attractive. A second factor will be availability, as while human coaches need to have time off, bots are available 24/7.      

Currently, data from various fields indicates that less than 0.5% of individuals prefer chatbots, and 50% of people report negative experiences with bots (2). In parallel disciplines such as telemedicine, studies reveal that users generally have less trust in their interactions with bots. The existing trust between humans and the bot is cognitive, as opposed to emotion-based as it is in human relationships (3).

In our view, bots are not ready to replace human coaches yet, or even soon, but has multiple ways it can be helpful which we have summarised from our recent research paper (4). 

These uses include assessment of client readiness for coaching and , features which many coaching platforms use, including EZRA, to enhance the coach-coachee chemistry, and ensure clients are motivated and ready to engage from the get-go.

The other aspect we have been exploring at EZRA is the role of AI as a tool to identify inter-sessional activities, these might include reflecting on aspects from the coaching session, readings or knowledge content or testing out new behaviours. It also includes helpful nudges that engage individuals to stick to the plans they have agreed to and reduce the prospects of being blown off course by events.

If we view AI as another tool that excels in specific tasks, we can harness its power now, while monitoring its development. When new use cases provide the evidence, they can positively contribute to the coaching process, as opposed to simply being fun tools to play with, we can consider integrating it more extensively.

Potential uses for AI in coaching

●        Assessment of client readiness for coaching

●        Coach-client matching

●        Recommendation of assignments

●        Supporting goal attainment between sessions through client nudges

●        AI coach supervision

●        AI chatbot delivered coaching

References

1) Passmore, J. & Tee, D. (2023a). The Billy Bragg Oxymoron: Assessing the powers of Artificial Intelligence in coaching conversations and knowledge synthesis.

2) Seitz, L.  Bekeier-Feuerhahn, S. & Gohil, K.  (2022). Can we trust a chatbot like a physician? A qualitative study on understanding the emergence of trust toward diagnostic chatbots, International Journal of Human Computer Interactions, 165

3) Weply (2021) Do you prefer chatbots over humans

4) Passmore, J & Tee, D. (2023b) Can chatbots replace human coaches? Issues and dilemmas for the coaching profession, coaching clients and for organisations, The Coaching Psychologist, 19(1), 47-54. doi: 10.53841/bpstcp.2023.19.1.47

Jonathan Passmore is a Professor of Coaching & Behavioral Change at Henley Business School, Senior Vice President at EZRA and Chair of EZRA’s Science Board