For the past few years, a lot of the conversation around AI and work has focused on learning the tools. Learn ChatGPT. Prompting, automation, agents… Learn whatever new platform appeared this week before another one arrives next Tuesday.
I understand the impulse. I use AI every day, and experimenting with these tools has changed the way I work. I’ve built GPT assistants, I’ve been creating Skills with Claude Cowork, and I use AI for research, writing, analysis, brainstorming, and increasingly for building things that I would have assumed required technical skills not very long ago.
But the more I use AI, the more I wonder whether knowing how to use it will be the differentiator we think it is. Eventually, almost everyone will know how to use AI. Maybe not at the same level or in the same ways, but opening an AI assistant, explaining what you need, uploading documents, or asking it to analyze something will probably become as ordinary as searching Google or creating a spreadsheet.
If that’s where we’re headed, then “I know how to use AI” won’t be much of a career advantage.
The more interesting question becomes: What can you do with it that someone else can’t?
The tool is becoming the easy part
When ChatGPT first appeared, knowing how to get useful results from it felt almost like a specialized skill. Prompt engineering became a term people suddenly added to their profiles. Courses appeared. People collected prompt libraries and formulas designed to produce better results.
Some of that knowledge was genuinely useful. But the tools themselves have also become much better.
You don’t necessarily need the perfectly engineered prompt you might have needed a few years ago. You can explain what you’re trying to accomplish conversationally, provide examples, upload documents, ask the AI to interview you before beginning, and correct it as you go. Increasingly, AI systems can work across multiple steps rather than just answering one isolated prompt at a time.
As that friction disappears, I think the value begins to move somewhere else. Knowing which button to press matters less when everyone can press it. Knowing why you’re pressing it, what you’re trying to accomplish, and whether the result is useful becomes much more important.
Your experience becomes part of the prompt
One thing I’ve noticed while building AI assistants is that the AI itself is only part of what makes them useful.
I can ask an AI to create a marketing campaign. So can thousands of other marketers. But I can also give it what I’ve learned from years of working in marketing and communications: how I think about audiences, what I look for in a customer journey, how sales and marketing should support each other, what makes an email sequence useful rather than simply automated, and what information I need before creating content.
Suddenly, I’m not just asking AI to “do marketing.” I’m giving it a way of thinking about marketing.
I’ve noticed this particularly while creating GPT assistants and Claude Skills. A surprising amount of the work involves taking something that previously existed mostly in my head and translating it into instructions, questions, criteria, examples, and processes that another system can follow.
And that has made me look at professional experience differently.
Years of experience aren’t just a list of jobs you’ve had. They’re also thousands of tiny judgments you’ve accumulated. You notice something that doesn’t look right. You know which question to ask next. You recognize when a technically correct answer doesn’t make sense for this particular situation. You know when something sounds impressive but probably won’t work.
AI can help us execute much faster. But we still have to bring something to the table.
Judgment might become more valuable, not less
There is an understandable fear that AI makes expertise less valuable because people can suddenly produce things that previously required specialized knowledge.
I’m experiencing a version of this myself. I’ve never been a programmer or developer, yet I’m now creating tools with AI. Someone who isn’t a designer can create surprisingly good visuals. Someone who isn’t a professional writer can produce a decent first draft. Barriers that used to be quite high are getting lower very quickly.
That’s exciting. It also creates an interesting problem.
When producing things becomes easier, we produce a lot more things.
There are more articles, posts, products, apps, automations, strategies, ideas, and experiments competing for our attention. Being able to make something is becoming less unusual. Being able to decide whether it’s worth making becomes much more important.
AI can generate twenty ideas in seconds. Which one is worth pursuing? It can draft a strategy. Does the strategy make sense? It can create a workflow. Should that workflow exist? It can write an email. Does the email understand the person receiving it?
It can build something.
Does anyone need it?
Those questions require something beyond knowing how to operate an AI tool. They require context, experience, curiosity, and judgment.
The advantage might be in the combination
I’ve spent much of my career as a generalist, moving between communications, content, email, CRM, strategy, project coordination, marketing, even accounting and admin support. For a long time, being a generalist could feel uncomfortable in a professional world that constantly encouraged people to specialize: pick your niche, become known for one thing, make your career easy to explain in a sentence.
AI has made me reconsider some of that.
When execution becomes easier, being able to connect different areas may become more useful. A marketer who understands sales can use AI differently from someone who only understands content. A writer who understands customer behavior can ask different questions. Someone who has worked across several disciplines can bring those perspectives together when solving a problem.
The AI might be the same. The person using it isn’t.
All of us bring different careers, industries, mistakes, successes, instincts, curiosities, and ways of seeing problems into the interaction. Perhaps that’s where part of our career advantage will come from: not from competing with AI on how quickly we can produce something, but from bringing everything we’ve learned into the way we direct it.
Knowing how to ask is only the beginning
We’ve talked a lot about prompting over the past few years, and asking good questions matters. But there’s another layer beyond prompting: you have to know what deserves a question in the first place.
You need enough understanding of a problem to recognize what’s missing. Enough curiosity to challenge the first answer. Enough experience to notice when something doesn’t quite fit. And enough confidence to tell the AI that its well-structured response is useless and start again.
This is one of the things I enjoy most about working with AI. The best sessions don’t feel like asking a machine for an answer and accepting whatever comes back. They feel more like working through something. I push, it responds, I disagree, we explore another direction, I provide more context, and occasionally something emerges that I hadn’t considered before.
The quality of that process depends partly on the AI, but it also depends on what I’m bringing to the conversation.
So what should we be learning?
There’s a big pressure right now to keep up. Every week brings another model, feature, platform, workflow, or prediction about which jobs are disappearing next. It can make professional development feel like an endless race.
I don’t think ignoring AI is a good strategy. Learning how to work with it is increasingly becoming part of knowing how to work. But I’m also becoming less convinced that we need to become experts in every tool that appears.
The tools will change. They will become easier, more capable, and more integrated into the software we already use. Some of the AI skills that seem sophisticated today will probably feel completely ordinary a few years from now.
So yes, learn AI. Experiment with it. Build things. Try the new tools. Figure out what they can do for your work.
But while doing that, keep developing everything you bring to them: your expertise, your judgment, your curiosity, your understanding of people, your ability to connect ideas, and your willingness to question an answer instead of accepting it because a machine delivered it confidently.
The career advantage in the age of AI may not belong to the person who knows the most AI tools.
It may belong to the person who knows how to make those tools more useful because of everything they already know.