The news: Generative AI (genAI) has been pitched as a path to efficiency. Instead, 95% of enterprise pilots have failed to show measurable impact, per MIT Media Lab’s The GenAI Divide. Productivity gains exist for individuals but aren’t scaling across organizations.
AI adoption among larger firms has swung sharply. Census Bureau data shows AI adoption among larger firms peaked at 13% to 14% in early 2024 and dropped to about 9% by mid-2025—a nearly 30% YoY drop. Adoption is climbing again but remains below the peak.
Poor output slows productivity: One reason for fluctuating adoption could be the surge of “workslop”—polished-looking AI output whose questionable quality slows teams down, per Harvard Business Review.
- Employees often spend more time fixing or rewriting flawed drafts than creating them from scratch, costing a company with 10,000 employees up to $9 million annually.
- The costs add up quickly. On average, workers lose two hours a day correcting AI misfires, with ripple effects across teams and management.
The problem is widespread: Roughly 40% of AI-generated output qualifies as workslop, according to ongoing research from Stanford University.
The issue is worsened by inadequate training: Although AI tools are widely accessible, only 13% of employees have received formal training, per HR Dive. That leaves the remaining 87% unprepared, more likely to misuse genAI, and inadvertently fueling workslop.
For marketing teams, the stakes are higher. A flawed AI draft for a campaign brief or customer email also risks damaging brand voice and credibility if errors slip through.
Possible remedies: Researchers suggest setting clear boundaries. GenAI excels at drafting, prototyping, and testing variants but shouldn’t replace human judgment in strategy, storytelling, or final copy. Businesses that draw this line—and invest in structured training—stand a better chance of reaping AI’s benefits.
Our take: AI adoption fluctuations serve as a warning. CMOs face a choice—either invest in structured AI training and workflow integration or risk eroding brand credibility and trust. Training employees to know when not to use AI will be as important as teaching them how to use it.