AI in approvals: will it speed up bureaucracy or make it merciless

Ars Technica examines the debate underway in US healthcare: will AI fix prior authorization — the procedure requiring an insurer to approve treatment before it starts? Optimists see algorithms approving obvious requests in seconds that today wait weeks for a human; skeptics see automation stamping out denials, with a machine's decision even harder for a patient to contest than a bureaucrat's.

Behind the industry dispute is a universal question — what exactly to automate in approvals. Experience across industries converges: AI is excellent at accelerating positive decisions against transparent criteria, but the right to refuse must stay with a human, otherwise the system optimizes toward denials — they're cheaper.

For a company in Kazakhstan this is a ready template for internal processes — approvals of payments, leave, discounts, contracts. Let an AI agent instantly pass requests that meet the rules and gather materials on contentious ones, but a rejection is always signed by a person, and every decision shows its author and reason. That way automation speeds up the work without becoming a wall your own employees and customers keep hitting.