AI Still Has to Wait for Approval

Give an organisation AI that can finish a task in minutes, and the result may still spend a week waiting for someone to approve it.

That gap helps explain why technical progress and social change move at different speeds. We can improve the technology while leaving the arrangements around it untouched.

Societal inertia sounds abstract. The reasons behind it are familiar.

A manager has authority because decisions pass through their department. A professional has spent twenty years building expertise. An organisation has paid for software, training and procedures that would be expensive to replace. Each has something to lose when the way we work changes.

Calling this resistance irrational is too easy. If your employer gets the savings and you get redundancy, being enthusiastic about the technology would require considerable generosity.

There’s also a coordination problem. A school can change how it teaches, but parents and universities may still judge it by exam results. A company can automate one department while the next department continues to demand the same forms. Changing your part helps only so much when everyone else depends on the old arrangement.

Some friction deserves to stay. People need ways to challenge decisions that affect their lives. Responsibility matters. An impressive demonstration tells us little about who will fix a mistake or compensate the person affected.

This will shape how AI develops and where it gets used. Institutions have reasons to favour systems that fit their existing rules and keep authority where it is. AI could make an inefficient process cheaper to maintain, giving its owners even less reason to replace it.

The result may be an uneven transition. Some activities will change quickly. Others will remain familiar long after the technology could support a different approach. Pressure can build until a competitor or a change in leadership makes reform unavoidable.

For anyone running an organisation, this deserves attention before the next AI purchase.

Pick one process you want to improve. Follow it from the initial request to the finished result. Look at where people wait, who can make decisions and which checks serve a useful purpose. Ask the people doing the work what they stand to gain or lose.

Then test a change to that process alongside the technology. Measure whether the whole job gets done better, including the checking and corrections.

Give the people affected a say in the redesign and a clear account of how the benefits will be shared. They need a reason to help make the change work.