Before I Knew Better

There is a saying in the startup world, often attributed to LinkedIn founder Reid Hoffman:

"If you're not embarrassed by the first version of your product, you've launched too late."

When I first heard that quote, I understood it in the context of business. Build something. Release it. Learn from it. Improve it. Don't wait for perfection because perfection is often just fear wearing a respectable disguise.

Eric Ries, author of The Lean Startup, takes the idea one step further. He argues that no matter how long you wait, you will probably still be embarrassed by your first version. Your standards will continue to rise faster than your ability to satisfy them.

At first glance, these ideas seem to belong exclusively to entrepreneurs. I have come to believe they apply just as much to writers.

Over the past several months I have been experimenting with artificial intelligence as a research, editorial, and organizational tool. Like many people, I began cautiously. It started with spelling and grammar. Then editing. Then structure. Eventually it became a genuine intellectual collaboration—not because the ideas ceased to be mine, but because the process of producing them changed dramatically.

Looking back now, I find myself embarrassed by some of the work I submitted for publication.

Not because I disagree with the arguments.

Not because I think the writing is poor.

Because I failed to disclose the role AI had played in helping me produce it.

The embarrassment goes beyond my own conscience.

It comes from wondering what my editors and colleagues must have thought as they read my work. Surely some of them recognized the cadence. Surely some of them suspected AI had been involved.

That realization forced me to confront a far more uncomfortable question than whether I should have disclosed its use.

What, exactly, makes someone a writer?

If AI is helping draft paragraphs, reorganize ideas, and accelerate the writing process, at what point does the author stop being the author?

For a time, that question unsettled me. It felt as though admitting I had relied on AI somehow diminished my claim to call myself a writer.

Ironically, wrestling with that question led me to the opposite conclusion.

Writing has never been merely the act of moving words across a page. It is the work of observing, synthesizing, questioning, judging, revising, and ultimately standing behind an idea. AI can accelerate many parts of that process, but it cannot assume responsibility for the conclusions. That responsibility remains mine.

Today, I no longer worry that using AI makes me less of a writer.

I worry only that failing to be transparent about its role made me less honest with my readers than I should have been.

At the time, I genuinely believed I had finished the work. I believed I had reached an ethical conclusion. I simply had not yet asked myself the right questions.

Today I would make a different decision.

Some people may see that admission as hypocrisy. I see it as evidence that experimentation works.

Had I refused to experiment because I feared making mistakes, I never would have encountered the ethical discomfort that eventually changed my mind. My current position did not emerge from abstract philosophy. It emerged from experience.

There is an old saying attributed to Voltaire that "the perfect is the enemy of the good." I think there is another enemy we discuss far less often: certainty.

If I had insisted on having every answer before using AI, I would never have discovered the questions that actually mattered.

The artist Leonardo da Vinci famously observed that art is never finished, only abandoned. Every finished work is really just a snapshot of what the artist understood at that moment in time.

Writing is no different.

When I submitted those early articles, I believed they were complete. They reflected what I understood then. As my understanding evolved, so did my standards.

Seth Godin has spent years encouraging creators to "ship" their work rather than endlessly polishing it in private. Austin Kleon argues that we should show our work before we feel ready because sharing accelerates learning.

Artificial intelligence has exposed another gap entirely.

Not simply between capability and ethics, but between production and reflection.

Technology allowed me to write faster than I had ever written before. It took much longer to understand what responsibilities accompanied that speed.

That understanding could not have arrived without first experiencing the technology itself.

Ironically, the very work that now embarrasses me is the work that made my current thinking possible.

Without those first imperfect attempts, I would never have developed the conviction I now hold: when AI has materially assisted the creation of an article, readers deserve to know. Not because the ideas cease to belong to the author, but because transparency is now part of authorship.

This realization has changed how I work.

Today, I believe AI disclosure should become as ordinary as acknowledging an editor, a translator, or a research assistant. The disclosure does not diminish authorship. It clarifies it.

Embarrassment, then, is not evidence that I failed.

It is evidence that I learned.

Perhaps that is the real lesson behind Reid Hoffman's famous quote. We should not be embarrassed because our early work was imperfect. We should be embarrassed only if years later we still believe it was perfect.

Growth leaves artifacts.

Sometimes those artifacts are awkward first drafts.

Sometimes they are businesses that almost failed.

Sometimes they are published essays that no longer reflect the fullness of what we believe.

I have all three.

And for perhaps the first time, I am grateful for every one of them.

Author's disclosure: This essay was developed with the assistance of AI as an editorial, organizational, and critical analysis tool. The ideas, arguments, final editorial judgments, and authorship are my own. AI was used to accelerate drafting, editing, and refinement while preserving my responsibility for the work's content and conclusions.