We tend to think the risk with AI is what happens when it replaces us.

But some new research points to another, less obvious risk: what happens when AI helps us.

*Researchers looking at more than 2,000 workplace tasks found that using AI to augment our work can create its own risks. In particular, when we become overly reliant on it, we may gradually lose some of the skills and oversight we still need.

What caught my attention wasn’t really the AI. It was the very human pattern underneath it.

We find something that works, so we keep doing it.

And because it works, we don’t necessarily stop to ask what else might be happening as a result.

The benefit is usually immediate and visible. The cost can be delayed and much harder to see.

I see this with leaders all the time.

A founder gets involved in everything because, in the early days, that’s exactly what the business needs. It works. Problems get solved. Decisions get made. Things move.

So they keep doing it.

Years later, they’re still jumping in. The problems are still getting solved, but their team has had fewer opportunities to learn to solve them without the founder. Now the founder wonders why nobody takes enough ownership.

Or the executive who builds a career by being the person who always delivers. Working harder works, so working harder becomes the answer to everything. Until the habit that helped create their success starts costing them the time and space to think, their relationships or simply the ability to switch off.

Or we avoid a difficult conversation because avoiding it keeps the peace. It works today. What we don’t see so easily is what happens to trust, honesty or the relationship when we keep avoiding it.

That’s why some patterns are so difficult to recognise.

The problem isn’t that they never worked.

It’s that they did.

Which makes me think the more useful question isn’t, “What bad habits do I need to stop?”

It’s this:

What am I still doing because it once worked brilliantly for me, but may now be costing me more than it’s giving me?

And perhaps an even harder question:

Am I still choosing this behaviour because it’s right for me now, or am I simply repeating it because it worked before?