The Impact of Government AI Usage on Democracy

A number of nations are exploring how artificial intelligence might assist with various tasks, ranging from tax processing to decisions about welfare benefits. Nonetheless, research indicates that citizens are not as optimistic as their governments, potentially jeopardizing democratic integrity.

“Focusing exclusively on immediate efficiency and appealing technologies could provoke public backlash and lead to a long-term erosion of trust and legitimacy in democratic systems,” states Alexander Utzke, at Ludwig Maximilian University in Munich, Germany.

Utzke and his team surveyed around 1,200 individuals in the UK to gauge their perceptions regarding whether human or AI management was preferable for government functions. These scenarios included handling tax returns, making welfare application decisions, and assessing whether a defendant should be granted bail.

Participants were divided; some learned only about AI’s potential to enhance governmental efficiency, while others were informed about both the advantages and the associated risks. The risks highlighted included the challenges in discerning how AI makes decisions, an increasing governmental reliance on AI that may be detrimental in the long run, and the absence of a straightforward method for citizens to challenge or modify AI determinations.

When participants became aware of these AI-related risks, there was a marked decline in their trust towards the government and an increased feeling of losing control. For instance, the percentage of those who felt government democratic control was diminishing rose from 45% to over 81% when scenarios depicted increasing governmental dependence on AI for specific functions.

After learning about the risks, the percentage of individuals expressing skepticism regarding government use of AI surged significantly. It jumped from under 20% in the baseline scenario to over 65% when participants were informed of both the benefits and risks of AI in the public sector.

Regardless of these findings, democratic governments assert that AI can be utilized responsibly to uphold public trust, according to Hannah Key de la Vallee from the Center for Democracy and Technology in Washington, DC. However, she notes that there have been few successful applications of AI in governance to date, with several instances of failures already observed, which can have serious consequences.

For instance, attempts by various US states to automate public interest claim processing have resulted in tens of thousands of individuals being incorrectly charged with fraud. Some affected individuals faced bankruptcy or lost their homes. “Mistakes made by the government can have significant, long-lasting repercussions,” warns Quay de la Vallee.