AI is everywhere in tech right now, and leadership roles are changing quickly because of it.
- Product managers are expected to spin up prototypes.
- Engineering managers are asked to manage more and more people with fewer and fewer resources, leaning on AI to make up the throughput.
- CEOs can ask questions directly against the analytics database.
- Engineers can go days without talking to each other and rubber-stamp code reviews because the LLM already reviewed it for them.
AI is not smart. It's artificial intelligence for a reason.
Knowing isn't understanding
I'm reading a book I recently found called He Who Fights with Monsters. One line speaks incredibly well to how AI is used in tech today:
"Knowing isn't the same as understanding. You've used a ritual magic skill book, but knowing practical applications isn't the same as grasping the theory."
The main character lives in an RPG-esque universe where he can level up his abilities and learn new skills through skill books. He doesn't have to read anything. He just accepts the book and receives the power, like Neo in The Matrix. But unlike Neo (or maybe just like Neo), knowing the skill doesn't mean he naturally knows how to apply it in every situation.
It's exactly the same for product managers, engineering managers, engineers, CEOs and everyone else using AI. It's so easy to snap our fingers and have a question answered, but you have to ask yourself:
- Am I asking the right question?
- Am I asking enough questions?
- Am I asking a question with a bias I can't see?
That product manager can spin up a prototype in a day, maybe a couple of hours, but that doesn't mean it's secure, performant or scalable. Those are questions engineers still have to ask, even when they're using AI. They have to know to ask, because AI is not smart. It's artificial.
An engineering manager who relies on summarized meeting notes to write reviews and understand what their direct reports need can easily believe everything is fine, when their star engineer is actually badly burnt out.
That CEO can point to the analytics database and say, "This feature we released without any marketing is blowing up." But because they're not a data scientist, they might not think to check whether overall platform usage went up at the same time.
Each of these people leveraged their way into a skill with AI, and each of them silently misled themselves. They went 10x faster at 20x the cost. WHY?!
Knowing where to tap
Sometimes the best decisions are made with 75% of the available information. But if you think you have 75% when you really only have 25%, you're no better than guessing. Much of that information is silently missing because the specialized skills that would have surfaced it are missing too. Some of those skills are no longer needed. Others vanish because we never knew they were there in the first place.
It's the story of the boat mechanic hired to fix a massive cruise ship. His rate was $100k. When he arrived, the crew described the problem and he walked around the ship for a bit. Then he pulled out his hammer, tapped a pipe a few times and said, "Your problem is fixed! I'll send you my invoice by email."
The client was shocked. "All you did was tap that pipe!"
The mechanic's response was simple: "The $100k isn't for the amount of time it took. It's for knowing where to tap, and which pipe."
As AI becomes more widely adopted, we start to lose those mechanics. We have engine rooms that no one has ever actually looked inside. Everyone got concerned when Outlook, then airlines, then major banks all started having outages. But GitHub has had a work-stopping outage probably once a week for the last two months. It's no surprise that the same kinds of flaws have finally reached industries that are much slower to roll out software updates.
Be the mechanic
So you want to be a leader in tech in this AI era? Figure out what those hidden skills are, and be that person.
Honestly, interviews aren't testing for that yet. So game the system, get the job, then show them what a great decision they made. Leverage your knowledge and experience. Be a 10xer with AI. Push yourself and stretch your boundaries, but acknowledge your own blind spots, and don't bring the ship down with blind confidence.