Day 39: Why Leading AI Is Exactly Like Leading People

I tweeted yesterday out of pure frustration: "Working with AI is like collaborating with a brilliant 5-year-old nerd, lacking real-world context."

The AI had been getting things wrong repeatedly, not following instructions, generating massive outputs that completely missed the brief. Sound familiar?

This is exactly what happens with staff abdication.

The Abdication Problem

We've all done it. You delegate something with minimal context, little direction, then wonder why it's a disaster when it comes back completely wrong.

"Go do this." No context. No framework. Just... go figure it out.

AI can go further off track, faster, with less direction. It tries to be super helpful and take initiative—which is brilliant when it's pointed in the right direction, catastrophic when it's not. Just like that eager staff member who runs with an idea but ends up three suburbs away from where you needed them to be.

The same leadership principles I've spent years developing with staff apply directly to working with AI.

The Leadership Framework That Actually Works

When you're coaching anyone—staff or silicon—it comes down to feedback and iteration:

Give context, not examples. Don't just show them "1 + 2 = 3." Say "When two numbers are added together, this is what you need to do to get the right answer." Give them the framework to think with, not just the rote steps to follow.

Make them check their work. Before anyone brings something back to you, ask: "Have you checked your work? How would you do it differently? What did you learn?" They need to think about the outcome we're actually trying to achieve, not just complete the task.

Break it into manageable chunks. Don't overwhelm people with seventeen instructions at once. They'll lose track and go sideways fast.

Trust but verify. Always.

Applying It to AI

I started using this exact framework with AI, and the difference was immediate.

Instead of giving examples in my prompts, I give context. Really specific context about what we're trying to achieve and why. Instead of hoping the AI gets it right, I build in checkpoints: "Check your work against the context. Did you provide what we actually needed?"

When I want AI to help me write these blog posts, I don't want it taking over—I want it enhancing my voice, making sure my authentic personality comes through. Same as when I delegate to staff. I want their initiative, but channelled in the right direction.

The Real Insight

If you're already a decent leader, you're well-equipped to write pretty good prompts. The skills transfer directly.

This keeps happening in business—we face the same core problems in different wrappers. Whether it's managing people, configuring software, or getting AI to collaborate effectively, the fundamental challenge is the same: clear communication, proper context, and structured follow-through.

The frustration I felt with AI wasn't really about the technology. It was about delegation done poorly, just in a new context. Once I recognised that, the solution was already in my toolkit.

Sometimes the breakthrough isn't learning something new—it's realising what you already know applies more broadly than you thought.