This is AI Frontiers. Today, Dutch government job cuts. The Dutch government has set a target to cut 25,000 full-time posts and save 3.5 billion euros a year before releasing any plan for artificial intelligence or AI. The savings were decided first, and AI was hired to explain them. On 26 October, the government announced savings rising to 3.5 billion euros a year from 2035. Yet the goals and savings plans for AI will not follow until the first quarter of 2027. Despite international headlines calling this an AI-led overhaul, the announcement is mostly organizational. It scraps or simplifies at least 500 rules a year, sets one desk for every two employees, sells surplus buildings, uses joint procurement, and cuts back advisory bodies. AI receives a single sentence promising forthcoming plans. Meanwhile, central government employment reached a record 160,000 at the end of 2025, up 41% on 2018. A 1.8% deficit is no emergency. The state secretary cited non-binding parliamentary motions calling for a 20% productivity gain through AI. Calling a reduction modernization costs fewer votes than calling it austerity. It also moves the burden of proof. The workforce now has to show it cannot be automated. Faster tasks do not automatically shrink payrolls. Time saved is not a post saved until someone shows where the time went. In Jabra's work life 2026 survey of more than 6,000 knowledge workers, 85% reported an increase in at least one area of work as saved time was reinvested. Only 8% of individual contributors are heavy AI users, against 30% of top managers. And 64% of contributors have no AI training or only a brief introduction, against 26% of top management. Before accepting a savings figure, ask three questions. First, what improves for the person receiving the service in processing times, completed cases, errors, and backlogs? Second, what evidence separates AI's contribution from other changes? Third, what does it cost for integration, data preparation, training, human review, and ongoing operation? When the Dutch plans arrive in 2027, the number that matters will not be 3.5 billion euros. It will be the processing time, error rate, or backlog showing what the saving costs the people who depend on the service. That's AI Frontiers.