“I can complete an entire whitepaper in two days now,” boasted my friend, a content marketer at an agency. He uses AI for research synthesis, argument structure, and clean professional prose. The whole package.
What took a week only takes two days now. A 60% reduction in time.
Except he published the same number of white papers this quarter as last quarter.
Why? Because, except for the writing process, the remaining bottlenecks were intact. Stakeholder alignment, approval cycles, decisions about what actually matters to the pipeline.
In 1987, economist Robert Solow made a famous observation: “You can see the computer age everywhere but in the productivity statistics.”
Nearly forty years later, we see AI everywhere but in the revenue numbers.
The Quality Debate Is Over
Today, AI-generated content is indistinguishable from human-written marketing content. Incredulous? Use the latest version of any genAI tool: it will banish any doubts you may have in this regard.
Ryan Law, Director of Content Marketing at Ahrefs, writes about how seven months ago, AI showed “sparks of brilliance” but needed significant human intervention. Today, with proper workflows and guardrails, AI can maintain consistent brand voice, follow complex editorial principles without fatigue, and self-evaluate its output. His conclusion: the articles sound the same and perform the same.
AI handles research, structure, and prose at a level that matches what skilled writers produce manually. The quality threshold has largely been crossed.
Given this, why aren’t companies seeing proportional productivity gains?
The Productivity Paradox, Again
Between 1970 and 1990, computing capacity in the US increased a hundredfold. Businesses poured billions into IT infrastructure. Yet productivity growth actually slowed, from over 3% in the 1960s to roughly 1% in the 1980s.
This became the Productivity Paradox. Dramatic advances in computer power coincided with declining productivity at both national and sector levels.
Were computers useless? Far from it. But productivity gains from transformative technology materialize slowly, with significant lag, measured in decades rather than quarters.
Productivity finally surged only in the late 1990s, nearly twenty years after initial IT investments.
We’re in that lag period now with AI.
The Workflow Problem Nobody’s Solving
Has AI changed content marketing? Content creation is now 3-5x faster, research synthesis takes hours instead of days, and first drafts have gone from decent to excellent.
But the time from brief to published content is roughly the same, strategic clarity about what to create still requires meetings, stakeholder alignment is still hard, and revenue impact remains unclear.
In short, AI has accelerated content creation without equivalent advancements on the other side of the value chain: the approval and coordination process.
Stuart Winter-Tear , who advises companies on AI product strategy, identifies three factors responsible for the bottleneck.
- Exception inflation. We can generate ten blog posts instead of three, but the approval process can’t handle ten — so we either produce three anyway, wasting capacity, or flood stakeholders with drafts and create coordination chaos.
- Coordination drag. Meetings multiply, “quick syncs” become the actual work, and the admin overhead doesn’t disappear — it moves into the seams between tasks.
- Quality debt. AI produces plausible-but-wrong content, and by the time someone catches it, it’s already been sent to five stakeholders and incorporated into two other documents.
With AI, content creation stopped being a bottleneck. The real bottlenecks now are strategic clarity, organizational alignment, and decision authority.
The Job That Emerged
The content marketer’s role is changing in ways that aren’t obvious from the outside.
Writing is now a smaller part of the job. The larger part is deciding which of ten possible narratives will actually resonate, navigating the gap between what product built and what sales can sell, knowing when to stop producing because the downstream process is already at capacity.
A content lead at a SaaS company described it well. She used AI to draft five variations of a product narrative in a single afternoon — each well-argued, each coherent. The writing was done. But that week’s actual work was something else entirely: two teams held fundamentally different views on what the product was, and who it was for. No draft could work until that disagreement was surfaced and resolved. Three days of conversations that had nothing to do with writing. The content followed in half a day once the alignment existed.
AI produced the drafts. It couldn’t see the organizational fault line underneath them.
This is translation work, not execution work. It requires organizational context AI doesn’t have yet — the political dynamics of who actually makes decisions, the unspoken strategic direction beneath official messaging, the stakeholder objections that matter versus the ones that don’t.
The Counterargument Worth Considering
The obvious challenge: if AI can match human writers, why won’t it eventually handle coordination and alignment too? It probably will, but over a long horizon.
Even then, there’s a difference between automating communication tasks and replacing organizational judgment.
Automating the memo doesn’t resolve the dysfunction that made the memo inadequate.
What the Computer Age Teaches Us
Companies didn’t just install computers and work faster. They redesigned processes, changed management practices, and restructured: slowly, unevenly, after painful adaptation.
New roles emerged that translated between technical capability and organizational reality.
AI will produce its own version of that shift.
The Lag Is the Story
The productivity paradox teaches us that transformative technology’s value is always underestimated in the short run and overestimated in the medium run.
The debate about whether AI can write as well as humans is largely settled. But it can’t make organizations faster, more aligned, or strategically clearer. It can’t resolve the coordination costs that dominate knowledge work. At least not yet, and not without the kind of organizational restructuring that takes years, not quarters.
Robert Solow saw the computer age everywhere but in the productivity statistics in 1987. In 2026, we see AI everywhere but in the revenue numbers.
History suggests both observations were correct. Just valid at different times.
The productivity will come. The question is what kind of work will still be recognizably human when it does, and whether we’ll be positioned to do it.
Musings by Madhav 
Leave a comment