AI Agents Overwhelming Public Services with New Requests

The emergence of artificial intelligence technologies has profoundly reshaped the landscape of public service interactions, leading to an unprecedented surge in applications and complaints worldwide. This phenomenon, termed "agentic flooding" by researcher Chris Schmitz, signifies a paradigm shift in how citizens engage with governmental and administrative bodies. As AI tools simplify the process of filling out forms and submitting grievances, public services across various countries are experiencing exponential growth in demand, necessitating a critical reevaluation of existing operational frameworks. This evolving dynamic presents both challenges and opportunities, urging institutions to adapt to an increasingly AI-driven administrative environment.

Schmitz's forthcoming paper, to be presented at the AI Ethics and Society conference, meticulously details 84 instances of this "agentic flooding" across 11 different jurisdictions. The data reveals a consistent trend: submissions remained relatively stable before 2022 but began a steep ascent with the wider adoption of AI, particularly tools like ChatGPT. For example, the United Kingdom's housing ombudsman witnessed complaints more than doubling, from 2,600 in 2022 to over 7,000 last year. Similarly, the United States' Consumer Financial Protection Bureau experienced a fivefold increase in complaints during the same period, with comparable surges in Brazilian judicial petitions and German parliamentary submissions.

A key aspect of this research highlights that the majority of these AI-assisted applications are legitimate, originating from individuals who are genuinely entitled to the services or benefits they seek. Unlike the "AI slop" observed in bug bounty programs, where AI-generated reports often lacked substance, the public service requests largely represent valid claims that might have otherwise gone unfiled due to the complexity or burden of traditional application processes. This underscores a critical insight: AI is not necessarily creating frivolous requests but rather empowering a broader segment of the population to navigate administrative hurdles that previously acted as barriers.

The underlying reason for this phenomenon, according to Schmitz, is the ever-increasing ease with which AI tools can be utilized. What once required precise prompting and significant effort with earlier AI models now often involves simple actions like pasting text or taking a photo with an AI-powered application to generate a comprehensive response. This accessibility has demystified bureaucratic processes, allowing more people to overcome the "administrative burden" – a policy term referring to the obstacles individuals face when attempting to access public services. The challenge for public services now lies in managing this amplified volume without corresponding increases in budget, a scenario reminiscent of how bug bounty programs were overwhelmed by reports without sufficient resources to vet them.

Looking ahead, Schmitz views this moment as a unique opportunity to reimagine and modernize social services for the AI era. While the task is substantial and largely unaddressed, the potential for AI to streamline interactions and make services more user-friendly offers a positive outlook. The experience of using AI to simplify tasks, such as tax returns, exemplifies the potential for a "good version" of AI-assisted public services. This period of rapid change prompts an imperative for institutions to fundamentally rethink their processes, leveraging AI not just as a tool for efficiency, but as a catalyst for more inclusive and accessible governance.