Replacing Public Servants with AI Could Come with Hidden Costs, Critics Warn

The government's strategy to reduce the public sector wage bill involves cutting jobs and increasing the use of technology, particularly artificial intelligence (AI). However, critics are raising concerns about the potential hidden costs associated with this approach. The government has announced plans to eliminate approximately 8,700 jobs across 40 core agencies, which it claims will save $2.4 billion over four years. Critics argue that the reliance on AI to replace knowledgeable bureaucrats may lead to unforeseen expenses that could undermine these savings. Roger May, a retired forestry consultant, expressed his concerns directly to Finance Minister Nicola Willis, stating that the transition to AI will require significant time and resources. He highlighted the lack of clarity regarding the costs and timeframes involved in implementing AI solutions, suggesting that these factors could erode the anticipated savings. The government has yet to provide detailed information on the rollout and licensing costs of AI, leaving many questions unanswered. Digitising Government Minister Paul Goldsmith acknowledged the complexity of the situation, noting that the costs associated with AI can vary widely. He emphasized the need for a coherent and centrally guided system to manage the diverse arrangements currently in place across government departments. However, the absence of a local AI provider capable of matching the scale of leading US models raises further concerns about the potential reliance on foreign technology. Experts like Professor Alexandra Andhov warn that the ongoing costs of enterprise-scale AI, including licensing fees and model upgrades, are often underestimated. The current pricing of AI solutions is heavily subsidized as companies compete for market share, meaning that the true costs may not be fully realized until later. Additionally, the evolving cybersecurity landscape poses further risks, as recent developments in AI have highlighted vulnerabilities that could impact public sector operations. The government must consider these factors carefully to avoid pitfalls in its AI implementation strategy.