Reflections on AI Coaching #4 – Will people use it?

My days as software engineer taught me that “build it and they will come” does not apply to technology, not always anyway. Just because we are able to create AI coaches does not mean people will use it. It is therefore crucial that we understand what factors influence users’ adoption of new technology, in this case AI Coaching.

The very first quantitative study we did using Coach Vici 1.0 in 2019 was to investigate what factors would influence users’ adoption of this type of technology. We asked users to have a conversation with Coach Vici and then complete the standardised UTAUT (technology adoption) survey that measures the role of performance, ease of use, social influence, infrastructure required and risk on the intent of users to use Coach Vici. We found that the most important predictor of using Coach Vici was the performance aspect. People were more inclined to use Coach Vici if they thought it was useful – more than any other aspect.  They paper is available here: https://journals.sagepub.com/doi/full/10.1177/21582440221096136

This finding made me realise that it is critical to create AI coaches that works, as obvious as it sounds. As we will see in a later post, even the early, rules-based versions of Coach Vici had remarkable efficacy. I foresee that the new version of Coach Vici that incorporates generative AI through ChatGPT will excel even more in the performance category and will therefore be even more appealing to users.

I also suspect that the social aspect (what do my friends think about using this technology) is enhanced through the significant media coverage of ChatGPT. Suddenly it is cool to use chatbots and not the struggle people often experienced in the past talking to a rules-based bot.

The challenge of course is to carefully and intentionally orchestrate ChatGPT inside an AI Coaching chatbot. This is where a thorough theoretical and scientific understanding of what coaching is, how it works and when to apply which approach is crucial. One of my past research papers for example shows that during career transitions, clients value certain coaching techniques far more than others: https://onlinelibrary.wiley.com/doi/full/10.1002/hrdq.21405?casa_token=VxEULdE43DYAAAAA%3Atxgc1XmxqTaozNWh4DAbuVs27kpPoKqGVGpCOEcpyIZILwgqbIKoqV-68mjlu-D-3-3WkZrklO_pIUiEHg

It seems that human-curated, evidence-based practice in crucial when creating AI coaches, especially since a badly created generative-based AI coach can hallucinate far more spectacularly than a human coach!