We are pleased to share an article entitled “So, Will Coaches be Replaced by Robots? No, but coaches will benefit from a coach-AI partnership” written by David Clutterbuck.
If you have experimented with any of the recent “intelligent” chat AIs, such as ChatGPT, you may well have been surprised at how rapidly they respond with information or questions that, at first sight, seem plausible. Then, when you look more closely, you see how inaccurate and random the responses are. But don’t be fooled – the technology is evolving rapidly and will get better quickly.
These chatbots work by accessing vast amounts of data and comparing with previous searches. They access just about any source on the internet. Earlier chatbots were simply mechanizations of diagnostic questionnaires. Their advantage was that they did most of what a human interviewer would do in a fraction of the time, effort and cost. But they couldn’t and can’t replicate functions like noticing a respondent’s reactions and following up a path of enquiry that’s not programmed in.
Artificial intelligence takes things further. The machine learns by creating its own algorithms, based on patterns it observes. And this is where it becomes dangerous for coaches. An AI in California has for several years been used alongside human therapists to treat trauma victims. A slight majority of the patients prefer the AI on the basis that they feel it is less judgmental!
Today, there are several organizations working to create AIs that will replicate coaching conversations. There are two approaches, which may be used in parallel. One relies upon gathering tens of thousands of recordings of coaching conversations and analyzing them. The challenge here is that there is no objective way to determine whether the coaching session was a “good” one, or how competent the coach was. (And even the best coaches have their off days!)
If you are signed up to one of the platforms that links coaches with clients and hosts the coaching session, you might well question whether an AI is “listening in” and what it is doing with the data gathered. If you suspect that the ultimate aim is to replace you with a robot, you will not be alone.
The second method starts with a specific coaching methodology and uses the experience of coaches steeped in that approach to create a library of questions and conversation structures. These can then be tested against a smaller (but nonetheless substantial) set of recordings of actual coaching conversations. An issue here is that coaches who stick to a single method or limited methods tend to score very poorly in coach assessment centers.
Neither of these methods takes account of what happens between coaching sessions, when much of the client’s deepest thinking often occurs. There is also the constant danger of introducing unintended bias, resulting from hidden patterns in the data the AI uses for its calculations. In general, it is not possible for an AI to explain how it reached a specific conclusion, because the bias is the result of many small decisions, rather than one decision that can be isolated.
So, who is at risk of being replaced? Our studies of coach maturity show that many beginner coaches have a fixed and repeatable model they follow – for example, GROW. As they mature, coaches go through a process of letting go of models, of processes, of the need for clear goals and of strictures against using themselves and their own experience. They also acquire a much more systemic perspective. It’s like the contrast between reading a joke out of a book and engaging in improv.
Today’s AI coaches can already equal the performance of many beginner coaches. They can follow a formula more consistently, without being distracted, don’t need to take copious notes and can fake compassion remarkably well. In short, they have the doing of coaching well in hand. It’s the being of coaching that is beyond the AI. As coaches mature, they start to integrate how they coach with who they are. They draw upon their own feelings and experience to shape and evolve the conversation. The key quality they bring that an AI can’t is their humanity.
Part of that humanity is their wisdom. Wisdom comes in three types:
✦ Skinny wisdom is all about being expert in a particular topic. An AI can know all there is within a narrow band of knowledge, but flounders when asked a question outside of its knowledge base. The new generation of AI is starting to overcome this limitation by integrating multiple sources of knowledge.
✦ Broad wisdom is what we gain from living and interacting with others. It covers ethics, love and many other qualities, and is the product of reflection on our experiences.
✦ Meta-wisdom combines many skinny wisdoms with broad wisdom to create judgment. Judgment involves emotional management, instinct and recognition of conceptual patterns – as does intuition. None of these are currently replicable by a machine.
I was recently introduced by a colleague to the reverse Turing Test. The Turing Test classifies a machine as intelligent if a human engaging with it in a detailed conversation is unaware that they are talking with a machine. The Reverse Turing Test suggests that, if a person cannot tell they are talking to a machine, they cannot be intelligent! Perhaps a better understanding might be that we need a new definition of intelligence…
Two conclusions arise from this analysis. One is that coach education needs to shift focus from teaching people to do coaching, to how to be a coach. Tools and techniques need to become the appendices in the book of coaching expertise, not the main text.
The other is that coaches need the skills of developing partnerships with the new technology, integrating the machine’s intelligence with their own intelligence and human qualities better to serve clients.
The core of all coaching is raising awareness by extracting clarity and purpose from complexity, in order to help clients exercise better judgment and achieve more positive outcomes. While both AI and the coach retain that perspective, the partnership that evolves will be of great benefit to coaches, clients and the wider human systems they serve.
Among the ways that the coach-AI partnership can assist the coaching process are:
✦ Making the coach more aware of what is happening in the room – for example, the hidden emotions and silent thoughts of the client.
✦ Revealing hidden patterns in the conversation and in the client’s thinking.
✦ Suggesting different questions and lines of inquiry.
✦ Helping the coach check their intuitions.
✦ Instantaneously making available information on relevant models and frameworks.
✦ Offering decision-making algorithms and frameworks.
✦ Supporting the coach’s post-session reflections with analyses of the conversation; and comparing these with other client sessions to extract themes of interest.
It will still be possible for coaches to survive and thrive without an AI partner. The human relationship is far more important than technical prowess. But coaches who lack personal maturity, and those who are reliant on a process or model for their practice, will have less and less to offer clients, compared to what the client can acquire from an AI. The growing issues of oversupply of coaches and commoditization of coaching may be resolved by forcing those coaches out of the market. Mature coaches who make effective use of themselves and their humanity will be the beneficiaries, as will their clients.
Tell us what you think about this article.
Let’s continue the conversation by connecting with your colleagues on our Facebook page
