The End to Governing in Prose (loosemore.com)

A recent discussion led by Sam Freedman highlights the growing concern among public servants about the inundation of claims and applications driven by AI-generated content within government services. The traditional declarative model used by public services—which relies on individuals exerting effort to submit claims as a proxy for sincerity—faces a significant threat as AI makes it easy to fabricate convincing submissions. Freedman suggests that a fundamental redesign of public services is necessary, proposing a shift towards a proactive model where the government anticipates and initiates support for citizens based on pre-existing data, rather than waiting for claims to be made. This shift, however, carries substantial risks. While proactive systems could enhance efficiency in service delivery, they may lead to codifying complex human situations into rigid rules, potentially compromising individual rights to appeal. The increasing capacity of large language models to generate realistic communication adds a layer of complexity to this issue. Freedman argues that as society navigates the challenges posed by AI, it may finally be prompted to undertake the long-overdue reforms needed to rebuild trust between citizens and the state, ensuring that public service design better addresses the needs of the most vulnerable individuals rather than simply taming demand through convoluted processes.