In 1930, John Maynard Keynes predicted that his grandchildren would work fifteen hours a week. Technology would do the rest. It is 2026. We work more than ever. Technology has not freed us from work. It has freed us from certain work and filled the empty hours with new work. Meetings about meetings. Reports nobody reads. Dashboards nobody looks at.
AI is about to do the same thing. Faster.
The savings paradox
The story always sounds the same. "With AI, we save twenty percent of the time we spend on reporting." The steering committee applauds. The team implements. And three months later, it turns out that twenty percent has been filled with more reporting. Or with extra checks on the AI output. Or with meetings to discuss the AI results.
The saving is real. The improvement is not. It is the same trap as with AI automation: faster is not the same as better.
This is the pattern we encounter in almost every organisation that deploys AI for efficiency. The focus is on speed. On throughput. On cost per unit. Those metrics go down. Everyone is pleased. But nobody asks the question that actually matters.
Has the work improved?
Simple question, difficult answer
Better is more subjective than faster. Faster you can measure with a stopwatch. Better requires talking to people.
Four criteria help. Ask the team each quarter: does the work you do now require more judgement than before? Are you building expertise? Do you have human contact in your work? Do you feel like you are contributing to something?
If AI saves ten hours a week and those ten hours fill up with more administration, with control work, with the overhead that remains once the interesting work has been automated, then you have shifted the work. Moved it from one pile to another, along a faster conveyor belt, but not improved it in any meaningful way.
The conversation nobody is having
Most managers talk about AI in terms of efficiency. Fewer FTEs, lower costs, faster throughput. Those are legitimate metrics. They are also incomplete.
There is a conversation that almost nobody is having. The conversation with the people who do the work. About how their role is changing. About what they are losing and what they are gaining.
Three sentences that help.
"Your role is changing. The parts that require judgement and creativity are becoming more important." This is not a reassuring phrase. It is an honest description of what happens when AI takes over routine work. The human becomes the evaluator, quality guardian, decision maker. That is a promotion, provided it is treated as one.
"I want to make sure you keep building those skills. Do this task manually for the coming period." This is the hardest sentence to say, because it sounds like a step backwards. It is. A deliberate, temporary step back to ensure the foundations remain solid. Without those foundations, the Junior Gap emerges: editors who have never learned to write.
"I don't just care whether the tools save time. I care whether the work that remains feels worthwhile." When a manager asks this and actually listens to the answer, something shifts in the team. People feel seen. That sounds soft. It is the foundation of every successful change.
What to measure
Always measure in pairs. Speed alongside quality. Efficiency alongside meaning. Cost savings alongside employee satisfaction. One-sided metrics lead to one-sided decisions.
And add a question that most dashboards miss: if we were about to switch off the AI tool, would users object? If the answer is no, you have built something efficient but irrelevant. If the answer is yes, you are on the right track. Also watch for AI drift: if nobody is looking critically at the output anymore, quality declines without anyone noticing.
Keynes was right about the technology. He was wrong about the outcome. The technology did exactly what he predicted: it made it possible to work less. But the choice to actually do so was left to the people. And the people chose more.
With AI, we get that choice again. The question is whether we answer it differently this time. Whether the hours that are freed up get filled with better work, or with more of the same.
That is a question you can only answer by asking it.