Why So Much AI Writing Feels Polished and Empty at the Same Time

Lee Domaszowec
5 days ago
5 min read

A few weeks ago I was serving on a committee tasked with reviewing and rating 500 grants over 10 days. It was exhausting. So many of those grants were clean, technically solid, but completely soulless.

There was nothing in it that suggested a real person had wrestled with the ideas, made choices, or felt accountable for the words on the page. You could have swapped in language from dozens of other organizations and nothing would have changed.

I couldn’t recommend any of those grants for funding, and other members of the review committee agreed.

That’s what a lot of nonprofit leaders haven’t fully absorbed yet: AI is an amazing tool, but without human input, it might hurt more than it helps.

Professional working remotely on a laptop, representing flexible strategic planning and digital operations for growing nonprofits and mission-driven organizations.

Why So Much AI Writing Ends Up Sounding the Same

Most leaders are just overwhelmed and looking for quick fixes. AI generated grants, emails, or social media posts CAN be really quick to generate. Perfecting the systems so that they sound good, however, takes time and expertise.

I know you’re short on time. You’ve got a donor update due Friday, a grant narrative due Monday, a board meeting midweek, and at least one program issue you didn’t plan for. AI looks like the one thing that buys you five minutes of breathing room, so you use it. You prompt, skim, clean it up a little, and move on. That makes sense.

The problem is that everyone else is doing the same thing.

They’re using the same tools, the same prompts, the same default structures. And AI doesn’t generate your voice. It generates an average. Over time, that average smooths out anything distinctive, specific, or human.

That’s how writing starts to feel interchangeable. Not wrong. Just empty.

And when writing feels empty, people don’t get angry or complain. They disengage quietly. They skim. They forget. They stop feeling connected to the person or organization behind the words.

Individual taking detailed notes in a notebook, reflecting intentional planning, research, and documentation for effective nonprofit strategy and program development.

What Most “Fix Your AI Writing” Advice Misses

There’s no shortage of advice circulating right now: Add quirks. Vary sentence length. Drop in a typo so it feels human.

I’ll be direct. That’s surface-level editing dressed up as strategy.

When I read those 500 grants, here’s what stood out most: the weakest writing wasn’t written by AI. It was written by people who never decided what they actually wanted to say before handing the work off to an AI model.

They didn’t define a stance.They didn’t choose a voice.They didn’t anchor the message in a real moment from their work.

When those decisions are missing, AI fills the gap with something safe and generic. That’s not a failure of the tool. It’s a failure upstream.

If you don’t define your voice, the model will supply one that sounds like everyone else.

Team member offering support during a group discussion, illustrating collaborative leadership, capacity building, and people-centered organizational management.

How Trust Actually Erodes

Trust rarely collapses in a dramatic moment. It thins.

A donor reads your newsletter and feels a little less connected than they did last quarter.A partner scans your website and doesn’t find language worth quoting.

A program officer reviews your outcomes, understands the numbers, but feels no pull toward the story behind them.

None of those moments feels decisive on its own. Together, they matter.

By the time leaders notice engagement slipping or support softening, the erosion has usually been happening for months.

Two professionals collaborating at a shared workspace, highlighting strategic consulting, team alignment, and sustainable growth planning for organizations.

How We Use AI at PhoenixFire Without Losing Our Voice

We use AI constantly. It saves time. It opens options. It helps us move faster. What it doesn’t do is speak for us.

Everything we write starts with a short, blunt brief. Four lines. No more.

Who’s speaking. Who’s listening. What we want them to do. One real piece of evidence that has to appear.

If we can’t answer those clearly, we don’t generate anything yet. That pause does more to protect voice than any clever prompt ever will.

Once the brief is solid, AI helps us explore structure or angles. But we always rewrite the opening ourselves. Tone is set immediately, and no model knows our clients, our work, or our stakes the way we do.

Specifics are non-negotiable.

“We help families in need” becomes “nineteen families stayed housed this quarter because rent assistance arrived before eviction.”“Our programs create community change” becomes “seventy-three neighbors walked into our pantry for the first time last winter after missing rent.”

If a sentence could describe any organization, it doesn’t belong to ours.

Before anything goes out, we read it out loud. Not silently. Out loud. If a line feels stiff in your mouth, it will feel stiff to a reader too.

And we protect texture. Real writing isn’t perfectly balanced. Thoughts run long, then tighten. Pauses matter. That unevenness signals that a human is thinking in real time. AI smooths those edges away by default. We put them back on purpose.

Close-up of handwritten notes during a work session, symbolizing structured thinking, goal setting, and operational clarity for nonprofit and small business leaders.

You Don’t Need a Communications Department. You Need a Habit.

Most leaders assume this problem requires more capacity. Another hire. Another vendor. Another tool.

It doesn’t.

What’s missing in most organizations isn’t effort or intelligence. It’s the habit of slowing down just enough to decide what you’re actually trying to say before you let a machine say it for you.

When writing feels generic, it’s rarely because the writer lacks skill. It’s because the thinking starts too late. The message never gets grounded. And once that happens, no amount of editing can put the humanity back in.

What actually works is simpler than people expect.

Before you generate anything, define the message in human terms. Who’s talking. Who’s listening. What you want to change on the other side of the page. That moment of clarity protects your voice more reliably than any technical fix.

From there, AI can support structure. It can organize ideas. It can surface options. What it shouldn’t do is decide how you sound.

That part still belongs to you.

As drafts come together, vague language is your early warning sign. Any phrase that could live comfortably on ten other websites probably doesn’t belong on yours. Replace it with something concrete. Something someone on your team actually experienced. Something a donor or program officer can picture without effort.

Over time, patterns emerge. Certain openings work. Certain phrases feel right. Certain rhythms sound unmistakably like you. Save those. They become your internal reference point.

That’s usually enough to stay ahead of the problem.

If You Want Help Tightening Your Voice

If you want help tightening your voice, send us one recent piece. A donor email, a social post, or a grant paragraph. We’ll show you where the voice drifts, explain why it happens, and send back a version that sounds like someone who stands behind their work.

Schedule a 20-minute consult at phoenixfiresc.com/consult and put “Keep it human” in the subject line.

Your voice matters. Treat it like the asset it is.