McKinsey Finds AI Makes People Faster. Companies See No Money. The Method Matters, Obviously.
McKinsey's latest State of AI report finds that most employees believe AI makes them more productive, yet relatively few organizations are seeing measurable financial returns. The report attributes this to companies layering AI onto existing processes rather than redesigning workflows. A task completed twice as fast does not increase revenue if surrounding systems remain unchanged.
This is a textbook demonstration of the productivity paradox, first documented by Robert Solow in 1987. The mechanism is dilution through stiction. Efficiency gains at the task level only convert to economic value when the surrounding organizational system is restructured to absorb the freed capacity. Faster work that produces the same output at the same cost is not productivity. It is theater.
McKinsey published the findings in their latest State of AI report. The study covers organizational adoption patterns and financial returns across surveyed companies.
- Pick one recurring weekly task you do faster now because of AI, such as drafting emails or summarizing documents. Write down exactly how many minutes you save per week. Expect a concrete number, probably 30 to 60 minutes.
- Identify what you actually do with those saved minutes. If the answer is 'nothing measurable,' you have personally replicated McKinsey's finding. Expect to realize the time evaporates into low-value filler.
- Redirect those saved minutes into one specific higher-value activity for one week, such as prospecting new clients or improving one existing deliverable. Measure whether that week produced anything you could attribute revenue or satisfaction to. Expect the exercise to clarify whether your AI usage is productivity or performance.