Public Services Brace for AI Agent Deluge

Artificial intelligence is profoundly reshaping how citizens interact with public services globally, leading to a significant surge in applications and complaints, a trend identified as “agentic flooding” by researcher Chris Schmitz. This phenomenon is evidenced by dramatic increases across various jurisdictions and service types, ranging from consumer complaints to judicial petitions.

In the United Kingdom, for instance, complaints directed to the housing ombudsman more than doubled, escalating from 2,600 in 2022 to over 7,000 last year, following the widespread introduction of AI tools like ChatGPT. Similarly, the United States’ Consumer Financial Protection Bureau (CFPB) experienced a five-fold growth in complaints during the same period. Brazil and Germany also reported similar substantial jumps in judicial and parliamentary petitions, respectively.

Chris Schmitz, who will present his findings at the upcoming AI Ethics and Society conference, has meticulously tracked this global surge. His research examines 84 distinct cases of potential agentic flooding across 11 jurisdictions, providing extensive evidence that AI tools are fundamentally altering the mechanisms through which individuals engage with public services. While his paper, for methodological reasons, refrains from definitively stating AI as the sole cause, the observed pattern is consistent: submissions were largely stable before 2022 but began to rise at an accelerating pace as AI technology became more accessible and diffused. Crucially, this growth shows no signs of slowing, indicating a potential for continued increases for years to come.

Schmitz attributes this escalating usage to the increasing sophistication and availability of AI tools. Tasks that once required considerable effort, such as gathering context and precise prompting with earlier AI models, can now be accomplished with remarkable ease, often by simply pasting text or capturing an image with an AI app to generate a comprehensive response. This ease of use empowers more people to engage with bureaucratic processes.

While this surge in volume presents a significant challenge for public services, similar to how bug-bounty services were flooded with low-quality, AI-generated reports last year, Schmitz draws a crucial distinction. Unlike bug-bounty programs which often received valueless submissions, the vast majority of new applications to public services are coming from genuine individuals with legitimate claims. Schmitz emphasizes, “The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing.” This suggests that many applicants were previously deterred by the administrative burden associated with filing claims – a concept known in policy circles as administrative burden.

For Schmitz, this era of agentic flooding, rather than being a problem of spam, represents a unique and valuable opportunity to reimagine and reform social services for the AI era. By leveraging AI to alleviate the administrative burden that previously prevented entitled individuals from accessing benefits, public services can become more accessible and efficient. This transformative task is substantial and largely unaddressed, yet Schmitz views it with optimism. He envisions a future where AI assists citizens with complex tasks, much like a helpful tool for tax returns, fostering a “good version” of public service interaction. He believes this could be the opportune moment to fundamentally rethink and redesign public service processes.