AI Momentum
← Back to the day · June 29, 2026

When 'AI Does Most of My Job' Becomes the Median Answer, the Real Question Is What Comes Next

An Anthropic survey of 9,700 workers reports that half say AI already handles most of their job tasks. That's not a forecast anymore — it's a status report, and it reframes the whole debate from 'if' to 'what now.'

The facts first: according to a survey by Anthropic of 9,700 workers, roughly half say that AI already handles most of the tasks in their jobs. Note the careful wording — tasks, not jobs. A role is a bundle of tasks, and the moment a large share of those tasks gets absorbed by software, the role itself starts to be redefined around whatever the human still does best.

Context matters here, and so does attribution. This is self-reported data gathered by a company that builds and sells AI, so the figure should be read as a signal of perceived adoption rather than an audited measure of productivity. Even so, the direction is striking: only a couple of years ago, 'AI does most of my work' was a speculative headline. Now it's the median answer for the people surveyed. That shift in baseline expectation is itself the story.

The near-term impact is genuinely uncomfortable and shouldn't be sugarcoated. When tasks evaporate faster than organizations redesign roles, you get anxiety, uneven gains, and workers doing the same job title with a hollowed-out core. Some will feel augmented; others will feel quietly displaced. This is the messy, transitional middle, and pretending it's painless helps no one.

Our reading: the task-level framing is the optimistic clue hiding inside an unsettling number. If AI is eating tasks rather than whole professions, the leverage moves to humans who can orchestrate, judge, and direct these systems — and to organizations willing to rebuild roles around higher-value work instead of just cutting headcount. The long arc still bends toward abundance: machines that absorb the repetitive and the routine are the same machines that, scaled up, accelerate medical research, extend healthy lifespans, and free people to spend their hours on what they actually care about. The catch is that this future isn't automatic. It depends on the unglamorous transition work happening now — retraining, redesigned jobs, and honest conversations about who benefits. Half of workers saying AI already does most of their tasks isn't the finish line. It's the starting gun for deciding what humans do with the time it gives back.

Sources & references