I’m sure we’ve all been there. It’s 5:30 p.m., you’re utterly exhausted, and you have to deliver tough performance feedback to a struggling team member. It’s so tempting to open an AI window and ask for what to say. Executives today are overloaded, navigating one crisis after another. Short on time, it’s no wonder that they seek quicker ways to get things done.
AI is the ultimate Fast Forward button. It has moved beyond writing code or summarising reports; it now offers to script our difficult conversations, performance reviews, and mentoring conversations. It promises an easy escape from the emotional weight of leadership.
Let me be clear: I am not anti-AI. When it comes to pure processing, analysing spreadsheets, or summarising two hundred pages of regulatory updates, AI is an indispensable management tool. You would be foolish not to use it. Mentoring is not an operational workflow; however, it is a profoundly human investment.
If you’re using AI to lead humans, it begs the question: if an algorithm tells you what to say to a struggling mentee, are you leading, or have you been reduced to a mouthpiece for a machine? And what will you do when the mentee goes off-script?
To understand why a script fails in those unpredictable moments, we first have to remind ourselves what we are actually trying to achieve.
What is Mentoring?
Let’s clear up a common misconception. While management is concerned with ensuring that tasks get done today, mentoring is developmental. It focuses on the growth of the person’s ability to handle tomorrow.
Professor David Clutterbuck argues that developmental mentoring centres on helping the mentee to develop their own quality of thinking, self-awareness, and resourcefulness. It is not a quick coffee chat, and it is definitely not just handing out answers.
Mentoring is a structured, two-way exchange of energy and vulnerability that happens entirely in the friction zone, that uncomfortable, high-tension space where growth happens. AI may act as a directive sponsor by providing an instant answer, but it cannot add the human context needed to help a mentee forge their own path.
Cognitive Theft: Answers vs. Questions
AI was engineered to remove friction and deliver solutions. In leadership development, however, friction is the point. A professional mentor asks questions based on training, experience and best practice, for the mentee to reflect on and answer themselves.
If a mentor spoon-feeds a mentee the solution, it robs the mentee of the cognitive struggle required to internalise the lesson. This is hard science, not my opinion. The foundational concept of the generation effect (Slamecka & Graf, 1978) established that humans internalise information far better when they actively generate it themselves rather than when they passively consume it.
This has been resoundingly validated by subsequent research. A massive meta-analysis spanning 86 studies confirmed the immense memory benefit of generating solutions over receiving them (Bertsch et al., 2007). We even know why it works: brain scans show that generating answers actively engages broad neural circuits across the brain, demanding deeper cognitive effort (Rosner et al., 2013)4.
When you use AI to hand your mentee the answer, you are committing cognitive theft. You might save ten minutes today, but you deprive them of building the internal resources and problem-solving muscle memory that they will need tomorrow.
The Decline of Intuition & Betrayal of Trust
I understand why leaders are tempted to outsource empathy. Having a deeply human, emotional conversation takes a toll, while AI offers a comfortable escape hatch from the fraught reality of human emotions. However, it does so at a heavy cost.
If you rely on an LLM to script your empathy, you stop practising the slower, fundamental skills of active listening and emotional regulation. You lose your leadership intuition – and ironically you also lose your edge over AI too. Conley (2024) holds that as AI absorbs more technical tasks, the new professional differentiator is no longer what you know (knowledge), but how you apply it (wisdom). “The future will not belong to those who simply know more. It will belong to those who can apply knowledge with judgment, context, timing, humanity and wisdom.” (Pin, 2026).
Imagine a newly promoted leader sitting across from you, vulnerably admitting they feel like an imposter. You respond with a perfectly formatted, AI-generated script. What do you think happens when they realise those polished words came from a bot? It is a catastrophic breach of trust. Consider instead the validation you could offer with the words: “I get it. I’ve been in your role, and it’s a lot.”
Empathy only carries weight because it costs the giver something: time, lived experience, or shared vulnerability. An AI response costs nothing and lacks context. Subconsciously, your people will feel that emptiness.
The reality is that organisations do need tech-savvy leaders (Ready et, 2020), and AI certainly does have a role in supporting both leaders and mentors. While it may sound counter-intuitive at first, successful integration of tech relies on psychological safety and human-to-human connection (Zaki, 2026).
Where AI Belongs in Leadership & in Mentoring
AI is brilliant when used strategically, and I am all for it in the correct context. A very recent Gartner report shows that sustainable returns on AI deployment stem from combining human capability with machine intelligence. The companies actually seeing ROI on their investment in AI are using tech to amplify human intellect, rather than to replace it.
How does that translate to mentoring? AI’s power is, of course, in automating administrative tasks in the background, perhaps matching mentees with mentors based on their profiles and requirements, or sending automated prompts between human mentoring sessions to help build and reinforce habits.
AI may also be used as a sparring partner. Leaders could use AI to role-play difficult conversations in a risk-free environment to help them prepare for unexpected responses. To clarify, sparring is useful to test your emotional regulation and explore different angles in a zero-stakes context, not for generating a script to memorise and recite to the mentee.
It goes without saying that any use of AI in a mentoring context is deeply confidential, so the necessary privacy safeguards must be ensured. Let’s assume that you’re using a closed-loop, enterprise-grade AI system, one where your prompts and data remain strictly within your organisation and do not train public models.
However, even within that secure perimeter, access must be strictly compartmentalised. Just as you would safeguard other HR records or sensitive data, you must ensure that the technical teams maintaining the AI platform do not have backdoor access to the highly sensitive vulnerabilities shared in a mentoring context.
The Mentor as a Human Anchor
Tech is definitely upturning how we work, but it cannot replace how we connect as humans. AI can provide support, but it is designed to speed up processes and smooth corners, whereas research proves that encoding learning experiences requires both friction and time. As AI takes over more work and knowledge becomes more accessible, the value of workers, leaders, and mentors will be in their wisdom, the human ability to understand the nuances of context and apply judgement and empathy.
As much as I favour AI and use it daily in processing tasks, it simply cannot act as the strategic and validating anchor a human being needs when they turn to a mentor.
I have a question for leaders: are you using AI to amplify your human leadership impact, or are you using it to shirk the responsibility of leading?
