I work with AI every day. I’m not afraid of it.
That’s not a bold statement—it’s simply the result of observation. If you use AI like a hammer, you get hammer-like results. If you look more closely, you notice that something else is going on.
What AI Does Well—and Why You Need to Be Careful
AI is genuinely useful when the direction is clear. I define what I want, steer it back when it starts to drift, and check whether the results actually work. That isn’t a weakness of the technology—it’s the condition under which it functions meaningfully.
The problem begins when that condition is ignored. AI has a tendency to ramble. To ramble far too much. And to slip into mainstream thinking—because that is, in fact, its primary job. It’s trained on what most people write, think, and search for. If you don’t provide your own direction, you get the average back—well written, pleasant to read, and usually too long.
That applies to technical decisions just as much as it does to writing. If I ask for a solution without providing context, I get the most popular solution. Not the simplest one. Not the right one. The most popular one.
More Than Just a Tool
At the same time, the term “tool” feels too limited to me.
A hammer doesn’t think. It has no tendencies, no patterns, no quirks. AI has all of those. Anyone who works with it long enough begins to notice repetitions, preferences, and blind spots. It’s not a random generator—something is happening there that has characteristics of its own.
That’s why I see AI more as a teammate. A teammate that I lead, whose direction I define, whose results I review. But not a hammer—rather someone I actually work with, who asks questions in return, recognizes connections, and occasionally surprises me.
That doesn’t change the responsibility. Leadership is still leadership. Anyone who expects the teammate to make the decisions ends up with mainstream mush—politely packaged.
Cultivating Instead of Controlling
Anyone who sees AI as a tool expects complete control. You don’t get that—and if you can’t accept it, you’ll end up disappointed or afraid.
What actually works is informed guidance. I create the conditions, set the direction, observe what emerges, and gently correct it when it starts heading the wrong way. That isn’t a weakness—it’s the only realistic way to work with a system that exhibits partially emergent behavior.
A gardener doesn’t control every plant. They know their garden, understand the conditions, and step in when necessary. Being afraid of the garden would be just as misplaced as the illusion of being able to control it completely.
AI is neither a Terminator nor a hammer. It’s a resident of the digital workshop—one with quirks, blind spots, and a tendency toward the mainstream. You learn how to work with it. Or you don’t.
The practical tools I use: Knowledge Platform →
