AI can help people finish individual tasks faster without making the whole organisation more productive. A University of Manchester report examines the gap, and argues that management, skills and redesigned workflows matter alongside the technology.
Watch Desk analysis
What happened
The report, “The AI Ambition Gap”, considers why wider adoption of AI has not yet delivered the productivity gains many organisations expected. It points to a timing mismatch: spending on retraining and workplace changes comes early, while any return may take longer. It also warns that speeding up one task does not automatically improve someone’s whole role or a business’s overall output.
The report cites examples of task-level improvements, including software developers completing around 26% more tasks with GitHub Copilot. But it also cites research finding coding activity rose by about 240% while software releases increased by around 30%, with human checking and testing still needed before code can ship. The Register reports that the study was produced by the University of Manchester’s Productivity Institute and commissioned by Bytes Technology Group.
The researchers argue that results depend on how work is organised and on the skills needed for judgement, coordination and oversight. They recommend strategic direction from management, stronger worker capabilities and AI governance, as well as choosing uses tied to customer value, operational bottlenecks or new products.
Why it matters
The distinction between a faster task and a more productive organisation is where many AI promises meet the working day. If the time saved is absorbed by checking, coordination or other tasks, a productivity gain on a demo does not necessarily become a gain on the balance sheet.
That makes the report useful for leaders weighing what to fund next: not just software, but the training and workflow changes that could help its benefits travel beyond one task. The cited figures are examples, not a guarantee that the same results will appear in every workplace.
Our read
The report’s strongest point is also its least glamorous: buying AI is not the same as changing work well. Organisations should ask what outcome they are trying to improve, and measure whether the improvement survives checking, hand-offs and the rest of the job. The chatbot is only one part of the process; the process is where the awkward questions live.
What to watch
- Whether organisations measure whole-workflow outcomes, rather than counting faster individual tasks.
- Whether training and redesigned processes accompany AI adoption.
- Whether future evidence shows task-level gains turning into sustained organisational productivity.
Discussion spark: Should organisations keep investing in AI when task-level gains are clear but whole-workplace productivity is not, or should they first prove the workflow changes pay off?
Sources and evidence
- AI projects not paying off? Keep at it, says Uni Productivity Institute (30 October 2026, 13:15 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.