Amazon corporate employees are pushing back against the company's internal AI mandate, telling The Guardian that the tools feel 'half-baked' and are actively making their jobs harder. The systems produce errors that workers must verify, correct, and cross-check with colleagues — turning what was marketed as a time-saver into additional work. One software developer put it bluntly: management keeps signaling that AI will make everyone faster, but neither they nor their peers actually feel it.
A large workforce analytics study now puts numbers behind what Amazon workers have been saying. ActivTrak analyzed activity data from 163,638 employees across 1,111 organizations over three years and found AI adoption correlated with increased workloads across every measured category — email volume up 104%, chat and messaging up 145%, business management tool usage up 94%. Its conclusion: 'AI is being used as an additional productivity layer, not a substitute for existing work.'
The data doesn't suggest AI is failing to speed anything up — in certain tasks, it clearly does. But that freed-up time isn't going back to workers as breathing room. It's being recaptured as higher output expectations. Former Google executive Mo Gawdat has described this as structural: efficiency tools don't reduce work, they raise the floor. The workload expands to fill the available capacity, and in a profit-driven system, management is the one setting the ceiling.
For the companies selling enterprise AI, the gap between what gets demoed and what workers actually experience day-to-day is turning into a credibility problem. ActivTrak's three years of data across more than a thousand organizations is harder to dismiss than individual complaints. If the productivity gains are real but workers feel worse off, that's not a messaging issue — it's a design one.