I Thought AI Was an Editing Tool. I Was Wrong.

A question changed the way I think about artificial intelligence.

Like many people, my first interaction with ChatGPT was cautious.

I wasn't looking for a collaborator. I certainly wasn't looking for a replacement for human thought. I was simply looking for a better spell checker.

For twenty years, I have worked full-time as a floral designer. My company, Atelier Ashley Flowers, has been part of my life for the last nine and a half years, but my relationship with flowers began long before that. My first job was at Chevy Chase Flowers, and for most of my adult life I never imagined doing anything else. Working with flowers wasn't simply my profession—it was what I wanted to do.

Until recently.

Over the last year and a half, I have also been teaching myself journalism—interviewing strangers, studying geopolitics, writing long-form essays, and slowly trying to find my voice as a writer.

When artificial intelligence exploded into the public consciousness, I approached it with curiosity but also a healthy degree of skepticism.

Eventually curiosity won.

At first, I used ChatGPT exactly as I suspect millions of people do: I pasted in paragraphs and asked it to tighten grammar, smooth transitions, and point out awkward sentences. It was useful, but it still felt like software. An exceptionally sophisticated version of Microsoft Word.

Then I began asking harder questions.

Instead of asking, "Can you fix this sentence?" I began asking, "Is my argument convincing?" Instead of requesting grammar edits, I asked it to challenge my assumptions, organize research, and identify weaknesses in my reasoning.

Somewhere along the way, the relationship changed.

Over the last several days, ChatGPT and I generated nearly fifty pages of original ideas. We explored authorship in the age of artificial intelligence, Walter Benjamin's concept of aura, social media as a digital panopticon, quantum observation, surveillance, creativity, and authenticity. I still don't know exactly where those ideas will lead, but I know I would not have reached them nearly as quickly on my own.

That realization led me to ask another question...

"If everyone is talking about taking Claude classes, using AI to write software, and automating entire businesses, what does that actually mean?"

The answer surprised me.

Most people, ChatGPT explained, use AI to answer questions.

A much smaller group uses AI to do work.

That distinction seems simple, but it may define the next decade of professional life.

Instead of asking an AI to summarize an article, businesses ask it to read hundreds of articles every morning, compare them, identify emerging trends, and deliver only the information that matters.

Instead of drafting one customer email, companies build systems that organize incoming messages, classify them, prepare responses, and notify a human only when judgment is required.

Even more surprising was learning that many people now use AI to write software without ever becoming traditional programmers. They describe the application they want in ordinary English. The AI writes the code. The human tests it, refines it, and decides what the software should accomplish. Increasingly, the human serves as the architect while the AI serves as the engineer.

It struck me that these people are not really taking "Claude classes."

They're learning how to think differently.

They're learning to see repetitive work as a system that can be designed rather than a task that must be completed manually.

When I considered my own work, the possibilities became obvious.

As a journalist, AI could monitor thousands of articles every day, organize research, prepare interview questions, compare drafts, and track submissions.

For my floral business, it could categorize customer inquiries, draft proposals, manage subscriptions, forecast inventory, and schedule social media.

For MetaFleurs, it could monitor developments in geopolitics, artificial intelligence, defense, and culture while building an archive that connects today's stories to ideas I explored months earlier.

None of those examples replace the work I actually care about.

They simply remove the work that prevents me from doing more of it.

The irony is that I began this journey believing AI was little more than an editing tool.

Today I think that view is far too small.

Artificial intelligence is not simply changing how we write.

It is changing how we organize knowledge, how we divide labor between humans and machines, and perhaps most importantly, how we think about thinking itself.

I still believe that human judgment, creativity, intuition, and lived experience remain irreplaceable. If anything, working with AI has made me value those qualities even more. The machine can help organize ideas, challenge assumptions, and accelerate execution, but it cannot decide what is worth caring about.

That responsibility still belongs to us.

For now, I have no ambition to become a software engineer.

But I do want to become fluent in this new language—not the language of programming, but the language of designing systems that allow people to spend less time on repetitive tasks and more time asking meaningful questions.

If my own experience is any indication, that may be where the real revolution begins.