The AI Ambition Gap

Almost every firm uses AI.

89 in 100 use AI in at least one business function. Adoption isn’t the problem.

Fewer than half get it past the pilot.

Only 46 in 100 have moved beyond AI pilots into everyday use.

Only a quarter changed the work.

About 1 in 4 redesigned their workflows around AI. The rest bolted it onto the processes they already had.

Just 6 get real value.

These “AI high performers” attribute at least 5% of their EBIT to AI.

But a task is only part of a job.

Judgement, coordination and the rest of the job become the constraint – and some of AI’s gain leaks away.

And a job is only part of a firm.

In an organisation, AI’s value depends on workflow redesign, management, skills and data. Without them, more of the gain leaks away.

Across the economy, much of it has leaked away.

Uneven diffusion, resource reallocation and institutions decide how much of AI’s gain survives. At every level, more of it leaks.

At the job level, a queue.

Coding is only part of a developer’s job. Reviewing and integrating the extra code becomes the constraint – completed projects rose only around 80%.

At the firm level, another queue.

Testing and release run across the organisation. Software releases rose just 30% – barely an eighth of the coding boost.

What decides it? Leadership.

AI sped up the task. Whether that gain survives the job and the firm is decided by leadership and process – who sets direction, and whether the work is redesigned around it.