In recent months, governments worldwide have experienced a surge in complaints, requests, and regulatory interactions linked to the widespread adoption of artificial intelligence (AI) tools such as ChatGPT. While AI was initially touted as a means to streamline public services and reduce costs, early evidence suggests it may be complicating administrative processes instead.

Former civil servant Sam Freedman highlighted a spike in case numbers across courts in England, Scotland, and Wales, where employment tribunal claims increased by 39% within a year. Analysts attribute part of this rise to claimants using AI to draft legal documents, some of which contain flawed or fabricated precedents, requiring judges and magistrates to spend additional time disentangling genuine grievances from AI-generated inaccuracies.

Within government departments, officials report being overwhelmed by an influx of AI-generated freedom of information (FOI) requests that strain resources. A similar pattern has emerged in members of parliament (MPs) offices, where automated letters sent on behalf of constituents place added burdens on staff tasked with ensuring no critical communications are overlooked. This flood raises concerns about the diminishing effectiveness of public consultations if they become saturated with low-value, machine-generated submissions.

The challenges extend to employment practices as well, with some employers noticing a rise in indiscriminate job applications disseminated more broadly than before. This response, facilitated by AI tools, makes it more difficult for recruiters to identify qualified candidates amidst the volume of generic submissions.

Although AI tools could theoretically help governments detect or filter mass automated communications, doing so risks neglecting vulnerable individuals who rely on such technologies to articulate their concerns. Constituents who improve their written interactions through AI still expect personal engagement from human representatives, limiting options for automated sorting in sensitive areas.

Experts note that the paradox of efficiency is not new: innovations like email promised greater convenience but gave rise to unintended consequences such as spam and information overload. Similarly, widespread AI adoption may lead to new bottlenecks as shortcuts once helpful become overwhelmed when universally applied. Analogies are drawn to urban traffic, where increased vehicle use has slowed travel speeds to those akin to horseback riders centuries ago.

Historically, systems have adapted through measures such as spam filters and congestion charges, and policymakers may need to consider introducing appropriate "frictions" to balance AI’s benefits with its challenges. For now, the growing complexity underscores that faster and easier processes in some public sectors often translate into increased workload elsewhere, suggesting that managing AI’s impact will require careful coordination rather than unchecked expansion.