The art of mentoring (yes, AI Included) 

Mentoring young developers has always been challenging. Finding that balance between teaching in a way that helps them grow, allowing them the space to figure things out and not just giving them the answers, is an ongoing balancing act.

However, if done right, it can be the most rewarding experience for everyone involved.

A new question I was recently faced with was, how do we adapt mentoring with the abundance of AI tool now available to us such as ChatGPT, CoPilot or Claude.

In my experience I don’t believe it is about fighting the tools but rather teaching people to think critically while embracing the change.

So how do we assist those entering the workforce with this natural dependence on AI?

Simple: we need to rethink mentoring for the AI age. And that starts with both sides understanding their role.

Mentoring is not a spectator sport; it is a relationship built on trust with both putting in the effort. Let’s have a look at how each party can make the most of the opportunity.

What can the mentee do to ensure success?

Show up prepared to sessions and ready to learn. Schedule the meetings to ensure both parties have set aside focus time to be present in the session and if possible, prepare specific questions on what you would like to ask. It is important for you to also take notes for reference at a later stage, if needed.

Honesty is crucial, no one is expecting you to understand everything when first starting out. Be sure to ask why. Hiding confusion can lead to issues later down the line and end up hindering growth.

It is important to communicate early when stuck on a problem. Learn the difference between what can be solved through trouble shooting, +- 30min and what is the start of a rabbit hole spiral that sets you back 3 days in silence.

Regarding AI usage, be transparent. There is nothing wrong with using AI, it is a useful tool. The problem comes when copying AI code, you don’t understand, into a project. If the solution has been crafted using AI and you don’t understand, consider starting the mentoring sessions with a statement like,

“Hey, I’ve used AI to come up with this solution. However, I don’t fully understand what it is doing? Can we go through it together?”

This approach allows for a learning opportunity and great collaboration.

How should the Mentor play their part?

Firstly, being a good developer does not automatically mean you will be a good mentor. Mentoring takes skill, patience and understanding of where your mentee is in their current journey.

With anything that is worth doing, this will take time. Ensure you show up consistently and block out real time to make a difference. Be present in the sessions.

When peer programming, verbally talk through what you are doing while solving the problem. It is a great way for your mentee to pick up your thought process and try emulating it.

Make your interactions a safe space. A mentee that feels safe to fail is more likely to try new things and take calculated risks. Your first code review will set the tone for this. You want to create an environment where questions and discussions are encouraged and not one where people are trying to hide their mistakes.

Always explain the ‘Why’, not just the ‘How’. Your job is to explain the why: “Why this chosen Architecture”, ‘Why this pattern”, “Why this business decision”. AI can tell them how to do something, but the why builds understanding and proper informed application.

Next is the modern dilemma, how to embrace AI, but still teach critical thinking. AI will be used whether you approve or not, that is the reality of the world we live in. So instead of fighting it, teach them to use it responsibly. Encourage them to look at the results critically and ask questions like:

“Is this actually a good suggestion?”
“Does this fit the code base?”
“Does it adhere to our best practices?”
“Is there possibly any security issue I’m not seeing?”

When starting a code review, lead with questions similar to:

“Walk me through what this does.”
“Why do you think the AI chose this approach?”
“Is there anything you would like to change?”

Lastly, avoid the curse of knowledge. Years of experience makes certain aspects of coding appear obvious, however, try to remember that you too were once fresh out of college, terrified, lost and React hooks made zero sense.

And here is something important: stay open to learning from your mentee too. They are coming in with fresh ideas, perspectives, exposure to new tooling and different ways of approaching problems. Some of my best learning moments have come from a graduate showing me new techniques I had never considered. Great mentoring is not just a one-way street – it is a conversation.

When in doubt: Be the Mentor you wish you had when starting out!

What should be avoided?

There are common mistakes, from both sides, that should be avoided to ensure the best outcome from this process.

From mentees:

Don’t treat your mentor like a search engine, they are not just a faster version of AI. Ask why and learn.

Copy-pasting without understanding will lead to a world of trouble. This applies to both traditional platforms, such as Stack Overflow, as well as AI.

Not asking for help early enough. There is not much anyone can do to assist when a deadline looms if they were not aware of the issue.

Asking the same question, repeatedly. Take notes and learn.

From mentors:

Don’t just give the answer, you are missing an opportunity to teach problem-solving.

Dismissing new tools, such as AI, without offering alternatives and guidance.

Not making yourself available and constantly cancelling meetings.

Avoiding difficult feedback.

The reality of AI

Let’s stop pretending that AI isn’t now commonplace in development. The question is not whether juniors will use it, they will. The question is whether we teach them to use it responsibly with understanding.

Your job as a mentor is to ensure they know the difference between good AI usage and bad.

Good being, using AI to explain why a hash map was used over an array in a certain instance. As opposed to bad being, copy-paste, ship and have no idea how it works when it breaks production.

Take the time to teach them that difference. Review the AI code together, show them what a good prompt looks like, help them identify when AI is confidently giving them the wrong answer.

 The bottom line

The goal of great mentoring is to build judgment, not just create another code pusher. It is about teaching someone to think through a problem and evaluate solutions critically. It is about growing a developer that will no longer need you looking over their shoulder.

It requires time, consistency and equal investment from both parties. The mentor needs to create the safe space for learning and genuinely want to see their mentee succeed. The mentee needs to be masters of their own destiny and own their own growth.

The best part, if you get it right, you are not only building better developers, but you are also building someone who will one day go on to mentor the next generation the same way you showed them.

Be the example they want to aspire to, not the cautionary tale they advise others to avoid.