The news: GenAI was pitched as relief from busywork—faster drafts, cleaner code, fewer routine tasks. An eight-month Harvard Business Review field study of a 200-person US tech firm revealed a more complex outcome. 

The study shows that AI adoption was mostly voluntary and employees leaned in even when AI use wasn’t mandated. The result: Work sped up, responsibilities multiplied, and AI-assisted tasks spilled into more hours.

Productivity rose, but so did work intensity. 

• Workers absorbed tasks once handled by other teams.
• Breaks between work tasks were filled with “quick prompts.”
• Multitasking multiplied as some employees managed parallel AI threads.

Why it’s worth watching: AI’s productivity gains are real—and broad. More than three-quarters (77.1%) of US full-time desk workers said AI tools make them more productive, per EisnerAmper.

Nearly half (46.5%) reported being somewhat more productive, and 30.6% say much more productive. Just 1.9% reported being less productive, while 20.4% saw no change. The upside cohort outweighs the downside by roughly 40 to 1.

That gap underscores how AI is changing the tempo of work. Quicker turnarounds raise the baseline for employee responsiveness and output expectations. 

• As tasks expand and deliverables increase, cognitive load rises, per Harvard Business Review. Because the extra effort is voluntary and often rewarding, strain can go unnoticed. 
• Workers are now managing both the job and the technology—learning tools, reviewing outputs, correcting errors—all while deadlines hold steady.

Implications for marketing teams: As AI adoption grows, sustainable gains will hinge on workflow orchestration, not just access to the latest models.

• CMOs should build pause points before AI launches, like a 24-hour hold between AI-generated campaign drafts and final approval to allow time for human review. 
• Batch AI work to protect focus, formalize peer reviews to avoid siloed outputs, and measure workload alongside output to minimize employee fatigue.

Teams redesigning workflows around AI will benefit over those that simply layer it on without adjusting output expectations.