The Cost of Waiting

Not adopting is also a decision, and its consequences are never recorded. An adoption has a sponsor and a budget line while a deferral has neither, risk registers hold hazards rather than benefits that failed to occur, and a director who blocked a pilot that would have succeeded is never questioned because the outcome is never observed. This paper sets the case for precaution, which is strong where harm would be severe and irreversible, against the evidence that regulatory caution carries measurable costs of its own. It reviews the first controlled evidence on AI-assisted productivity and attends less to the magnitudes than to the distribution, since gains falling largest on the least experienced make timing an ethical question rather than only a commercial one. A two-sided framework follows, asking severity, reversibility, time-sensitivity, distribution and evidence threshold of deploying and of waiting alike, with demanded evidence matched to how hard each choice is to undo. The argument is for symmetry rather than speed, and it accepts that some organisations should not adopt at all for some purposes.