NBER Survey: AI Adoption Barely Moves Jobs and Productivity Metrics

AI is supposed to be the great work bulldozer. Flatten the boring tasks, squeeze out waste, spit out higher output with fewer people. So when I see a big survey saying the real-world impact on jobs is basically zero and productivity is basically a rounding error, my first reaction isn’t “wow, surprising.” It’s: of course. We bought the power tools and then used them to rearrange the garage. Based on public reporting, an international survey of nearly 6,000 CEOs and executives found that AI adoption is already widespread, but the average employment effect over the past three years is close to 0.0%, and productivity is only up a little, around +0.29%. It’s being framed as another version of that old “productivity paradox” idea: everyone swears the tech is everywhere, yet the numbers barely move. That’s either a reality check… or a warning sign. For content creators and marketers, this lands right on the nerve. Because this is the one area where AI looks like it should show up fast. You can spin up an ai content creation tool in minutes. You can run an ai writing tool like a slot machine: headline, outline, intro, rewrite, shorten, expand, translate. There are ai content generator products that can fill a calendar before lunch. If any function should show “more output,” it’s content. But here’s the uncomfortable part: more content is not the same as more value. Most teams don’t have a content problem. They have a “content that actually moves someone” problem. Imagine you’re a marketing manager. You get budget for a new content marketing ai tool. You plug in a marketing content generator ai, and suddenly your team can produce five times as many posts. Cool. Now what? Your distribution doesn’t magically expand. Your audience doesn’t magically trust you. Your product doesn’t magically become easier to understand. So you publish more. The internet shrugs. Your boss asks why leads didn’t go up. And your team spends the “saved” time tweaking prompts, editing bland drafts, and arguing about brand voice. The work doesn’t disappear. It just changes shape. That’s how you get “widespread adoption” with “minimal impact.” The tech slips into the workflow, but the bottleneck is still human: taste, judgment, positioning, and knowing what to say that isn’t already said. A lot of AI use right now is basically faster first drafts. An ai writer can help you get unstuck, sure. But the draft isn’t the hard part. The hard part is deciding what’s worth drafting in the first place, what you’re willing to stand behind, and what you can prove with real examples. If your strategy is fuzzy, AI will happily scale the fuzz. This is why I don’t buy the casual promise that “AI will boost productivity” like it’s automatic. Productivity is not “typing faster.” Productivity is “getting results with less waste.” If you use an ai content creator tool to make 30 pieces nobody reads, you didn’t get more productive—you just got louder. And the job angle matters too. People keep predicting a content bloodbath: fewer writers, fewer junior marketers, fewer agencies. Yet this survey suggests jobs haven’t moved much. That could mean AI isn’t replacing people as fast as the loudest voices claim. Or it could mean companies are doing the classic thing: they add tools, keep headcount, and expect “more” without changing how decisions get made. I see a more subtle risk for creators: the work splits in two. On one side, commodity content becomes cheap and fast, powered by content creation software ai and an ai content automation tool. On the other side, the “human” layer—original reporting, sharp POV, creative direction, and real relationships—gets more valuable and more rare. If you’re stuck in the middle, doing generic blog posts that could be written by anyone, AI doesn’t just threaten your job. It pressures your identity. Now, there’s a fair counterpoint: three years is not a long time. Big shifts take time to show up in productivity data, and companies are still learning how to use these tools. If all you did was bolt an ai content workflow tool onto a messy process, of course the results are small. A content intelligence platform or content research tool might not pay off until the team changes how it plans, measures, and ships work. Maybe the “real” gains show up later, once companies stop treating AI like a toy and start redesigning the system around it. But that’s exactly my point: the gains aren’t inside the tool. They’re inside the choices people make with it. Picture a lean team that uses an ai content marketing platform responsibly. They don’t use it to flood channels. They use it as a content ideation tool, a content idea generator, and a second brain to pressure-test angles. They use it to repurpose strong ideas, not invent fake ones. They keep humans accountable for truth, tone, and taste. That team might genuinely get faster without getting worse. Now picture the opposite team. They use the ai content generator to fill space. They let a tool decide what’s “important” because it sounds confident. They stop talking to customers because the tool can “research.” They ship more, learn less, and slowly lose the ability to tell what good looks like. That team might look productive in a spreadsheet while the brand quietly rots. So when executives say they’re optimistic about future boosts, I’m not rolling my eyes. I’m asking: optimistic about what, exactly? More output, or better outcomes? Because if all we do is automate mediocrity, we’ll get a world where the average gets cheaper—and the exceptional gets harder to find. If AI is already everywhere but the measurable impact is tiny, is that because the tech is overhyped, or because most companies are using it in ways that can’t possibly move the needle?