The promise: AI will make your job easier. The reality: AI means your boss can track everything you do and expects you to do twice as much. A new study validates what Amazon employees have been saying - and the implications extend far beyond one company.
Amazon corporate workers told The Guardian that mandatory AI tools are counterproductive. The systems make frequent errors requiring verification and correction, ultimately increasing time per task. One software developer put it bluntly: "I and many of my colleagues don't feel that it actually makes us that much faster."
Now we have data backing up those complaints. Workforce analytics firm ActivTrak analyzed activity across 163,638 employees in 1,111 organizations over three years. The results are damning: AI adoption increased workloads across every measured category. Email volume up 104%. Chat and messaging up 145%. Business management tools up 94%.
The researchers' conclusion? "AI does not reduce workloads." Instead of replacing existing work, AI functions as "an additional productivity layer" - a polite way of saying it enables companies to extract more output while employees bear increased burden. Any time freed by AI efficiency gets immediately filled with additional tasks.
This is the automation paradox playing out in real time. The same pattern repeated with email, mobile phones, and Slack - tools promised to make work easier instead just raised the baseline of expected productivity. Technology amplifies existing values, and in capitalism's case, that means maximizing extraction rather than improving worker wellbeing.
What makes Amazon's AI push particularly concerning is the combination of surveillance and productivity pressure. When management can measure everything and AI theoretically makes you faster, the expectations ratchet up accordingly. The technology doesn't give you time back - it just resets the bar higher.
One former Google executive noted that mobile phones were advertised as work-reducers but intensified work culture instead. AI is following the same script, except with more sophisticated monitoring and higher stakes. When your productivity is measured algorithmically and compared against AI-augmented peers, there's no such thing as 'good enough' anymore.
