The Inflation-Deflation Trap: When AI Automates Both Sides of Communication

I watched a client's team spend forty-five minutes drafting an email last month. Strategic vendor relationship, delicate ask, lots of stakeholders to satisfy. They used AI to expand their three core points into eight polished paragraphs.

The recipient's AI summarized it back to three bullet points.

Nobody wrote. Nobody read. But everyone felt productive.

The Job Content Gets Hired For

I've leaned on Clayton Christensen's milkshake framework to explain product strategy more times than I can count. The insight: people don't buy products, they hire them for a job. The milkshake got hired for the boring morning commute — thick enough to last, one hand on the wheel, keeps you full till lunch. One product, one clear job.

Content isn't a milkshake. It's the only thing I know that gets hired for two opposing jobs simultaneously.

The writer hires it to expand a thought — to add nuance, build credibility, show they've thought it through. The reader hires it to compress that thought back down — to extract the decision, the insight, the thing they actually need. Inflate and deflate. Same artifact, pulling in opposite directions.

For all of professional communication history, that tension was the craft. You expanded your thinking into writing that justified a reader's time investment. The skill was making the expansion worth the compression cost.

AI just automated both ends.

Theater in the Middle

There's a Marketoonist cartoon making the rounds that nails it: AI turns my one bullet point into a long email I can pretend I wrote, then your AI turns my long email back into one bullet point you can pretend you read.

Funny because it's true. Important because of what it reveals.

We've built an elaborate performance where the actual human-to-human transfer never happens. I start with "Need approval for Q3 budget increase," let AI inflate it into eight paragraphs about strategic alignment and market conditions, ship it across the org, and you feed it into your summarization tool that spits out "Wants more budget for Q3."

The thing in the middle? Theater. We're burning tokens — and credibility — on a round trip that adds nothing.

I've survived enough technology cycles to recognize this pattern. Remember when PowerPoint let everyone become a designer? Suddenly every update needed forty slides with animations and custom templates. The expansion was free, so we expanded everything. Most of those decks could've been an email. Most of those emails could've been a Slack message. When the cost of expansion drops to zero, we expand everything — whether it needs it or not.

The Round-Trip Test

Here's the uncomfortable question I'm sitting with: if a one-line summary does the same job as the whole thing, was the whole thing ever the job?

Not "can AI write it?" — obviously it can. Not "can AI summarize it?" — obviously it can. The test is whether what you're making has a job that survives the round trip.

I'm working with audit teams navigating this right now. Traditional audit communication was built on documentation as evidence — the expansion justified the conclusion. You showed your work. The length was part of the job: demonstrating rigor, creating a defensible trail, building stakeholder confidence.

Now partners are asking: if the AI summary captures the finding, why does the full report matter? And that's the right question, but it reveals something most people aren't ready to hear.

Some of our expansion was always theater. Not all — but some. The extra paragraphs that sounded professional but added no new information. The hedging and qualifications that protected us legally but communicated nothing. The formal structure that signaled seriousness without creating clarity.

AI just made that waste visible.

What Survives Compression

So what jobs actually survive the round trip?

A decision that requires trust. If I'm asking you to approve a career-altering investment, the bullet point states the ask but doesn't build the confidence to say yes. The expansion isn't filler — it's the evidence that lets you trust the recommendation enough to act. That's a job AI can assist but can't replace.

A relationship that compounds over time. The partner who writes the client email themselves, in their own voice, with the specific detail that shows they remember last quarter's conversation? That's not expanding a bullet point. That's investing in a relationship where the medium carries meaning the message can't. Your AI can't fake having been in the room.

A moment of genuine insight. When someone shows you how two seemingly unrelated things connect, and your mental model shifts? That survives compression because the value isn't in the words — it's in the reorganization that happens in your head. The summary might capture the conclusion, but it can't recreate the journey that makes the conclusion stick.

Here's what doesn't survive: status updates, progress reports, most meeting recaps, and about 60% of the "strategic overview" decks I see. If the summary does the job, the job was never the expansion.

The Audit You Should Run Monday

I'm not arguing against AI tools. I use them. But I am arguing we need to get honest about what we're making and why.

Before you inflate that next bullet point into a document, ask: what job is this getting hired for?

If the job is "document a decision" — send the bullet. If the job is "cover my ass with a paper trail" — you've got bigger problems than AI. If the job is "build trust for a high-stakes ask" — then write it yourself, because the fact that you wrote it is part of what does the job.

The companies winning right now aren't the ones using AI to write more. They're the ones using AI to write less — automating the theater so humans can focus on the communication that actually builds trust, changes minds, or deepens relationships.

I'm watching accounting firms struggle with this in real time. Junior staff used to spend hours expanding partner notes into client-ready reports. AI does that in seconds now. Some firms are celebrating the efficiency. The smarter ones are asking: if expansion was the junior's job, what job do we hire them for now?

That's not a comfortable question. But it's the right one.

What This Means for Your Work

Here's what I'm telling clients: run the round-trip test on your last five important communications.

Take the document you sent. Feed it to an AI summarizer. Compare the summary to what you wish the recipient understood.

If they match? You've been creating theater. Stop. Send the bullet.

If they don't match? You've found the job that survives compression. That's the only thing worth making. Everything else is just burning credibility in the middle.

The craft isn't dead. It's just that the craft is no longer expanding — it's knowing what jobs require expansion at all.

Your move: Pull up your sent folder. Find the last "important" email you wrote. Run it through ChatGPT or Claude with the prompt "summarize in one sentence." If that summary would have done the job, you've found your first piece of theater to eliminate.

What job is your content actually getting hired for? That's the question that separates the communication that matters from the performance in the middle.