Give two people access to the same AI tools and ask what they want to do with them.
One might say, “I want to finish this report faster.”
The other might say, “I want to build something so we don’t have to keep making this report.”
Both are reasonable answers. But they point toward different ambitions: getting through the work in front of you, and changing what you’re capable of doing next.
That second possibility is what interests me most about AI.
I think of it as an Iron Man suit. Not a replacement brain.
The point of the suit is that a person can do things they couldn’t do without it. It extends their reach. It doesn’t remove the need to decide where to go, what matters when they get there, or what consequences they’re willing to accept.
That’s the relationship I want with AI: more capability, without surrendering judgment.
Helping is the starting point. Building is the opportunity.
There is real value in using AI to draft an email, summarize a meeting or help make sense of a spreadsheet. Saving an hour is useful. Not every task needs to become a software project, and nobody needs to invent extra work to justify the time they save.
But I don’t want our imagination to stop at finishing the same tasks a little faster.
Consider a hypothetical purchasing manager. They could use AI to summarize a supplier’s quote. Or they could use it to help build a repeatable comparison that flags changes in price, lead time and specifications—and shows where a person needs to make a call.
A chef could ask for menu ideas. Or use AI to help turn their own experience into a planning tool that compares portions, preparation time and ingredient costs before a service.
Neither example requires the person to stop being a purchasing manager or a chef. Their knowledge is the starting point. The tool gives them another way to put it to work.
And “building” doesn’t have to mean writing code. It might mean a clearer process, a useful training resource, a prototype or a way to answer a recurring customer question reliably.
The question changes from “What can AI finish for me?” to “What could I make possible with it?”
A finished answer is not the same as a finished thought.
There is another way to use these tools: accept the answer because it sounds complete.
That is the distinction I care about more than who uses AI frequently or who writes the cleverest prompt. Are we using it to extend our thinking, or to avoid engaging with the problem?
Imagine a polished recommendation to change a supplier. The writing is clear. The comparison looks organized. But someone still has to ask whether the quotes cover the same specifications, whether the delivery dates are realistic, and whether the cheaper option creates another cost elsewhere.
Those questions don’t disappear because the document looks finished.
AI can help investigate them. It can suggest a consideration you missed, test a competing explanation or expose an inconsistency. Keeping your judgment doesn’t mean assuming you know better than the model. It means being willing to challenge both its answer and your own starting position.
What I would not accept is “the AI said so” as the end of the explanation.
Delegate the work. Keep ownership of the result.
I don’t think responsible AI use means doing everything manually and asking a machine to watch. That would miss the opportunity.
Delegate the repetitive work. Let it produce a first version. Use it to explore approaches you wouldn’t have had the time or skills to try alone.
But know what the result is supposed to accomplish, what it depends on, and how you’ll tell whether it works.
For a rough idea, that check might be a conversation. For a customer-facing process, it should include testing real cases. For a consequential financial decision, it needs a much higher standard of evidence and review.
You don’t need to understand every line of generated code to own a useful tool. You do need a credible way to validate its behavior, recognize its limits and get help where your expertise ends.
That is ownership. Not doing every step yourself, but refusing to let responsibility disappear between the prompt and the output.
Build your judgment along with the tool.
The opportunity I find most exciting is for someone who understands a problem but previously lacked a practical way to build a solution.
AI can help them make a first version. Then the important work becomes concrete: try it, find the weakness, change it, try again.
That process is also a chance to learn. Ask why an approach was suggested. Ask what would make it fail. Compare an alternative. Look at what happened when real people used it.
That is a very different ambition from generating more documents nobody questions. I would rather have one useful process someone understands and improves than a folder full of impressive-looking answers nobody owns.
The suit is not the achievement. What you choose to do with it is.
Use AI to reach further. To make something you couldn’t make before. To ask a harder question and stay with it long enough to find a better answer.
Wear the suit. Keep your judgment. Build something that matters.