THE HAGUE--The growing use of artificial intelligence by politicians and government institutions could fundamentally change how democratic decisions are made, potentially shifting influence away from elected representatives toward technology companies and the algorithms they control.
That warning comes from Dutch researchers Marcel Boogers and Lucyenne van de Coevering, who argue that the increasing reliance on AI in government raises concerns that go far beyond inaccurate information, automated speeches or politicians using chatbots to prepare parliamentary questions.
In an analysis published Wednesday by Dutch public administration platform iBestuur, the researchers contend that the real danger lies in allowing artificial intelligence to influence how competing public interests are evaluated and which priorities ultimately shape government policy.
Artificial intelligence is increasingly being used by politicians and civil servants to summarize lengthy documents, analyze budgets, prepare legislative proposals, draft policy papers and process public feedback.
While these applications can improve efficiency and give elected representatives faster access to information, the researchers warn that they can also introduce new forms of dependence that are difficult for the public to recognize.
Who decides what information politicians receive?
One major concern is the growing role algorithms play in selecting and presenting information.
When AI summarizes a government report, analyzes a budget or prepares policy recommendations, it determines which information receives attention, which arguments appear most relevant and which issues may receive less emphasis.
These decisions can influence how politicians understand problems before they even begin debating possible solutions.
The researchers question who ultimately determines how information is filtered, what assumptions guide the analysis and whether certain viewpoints are systematically favored over others.
This becomes particularly important when elected representatives rely on AI-generated summaries instead of examining the original documents themselves.
Democracy involves more than finding the fastest solution
The researchers emphasize that democratic government is not simply about identifying the most efficient answer to a problem.
Political decisions frequently involve competing interests that cannot be resolved through calculations alone.
Decisions about housing, taxation, healthcare, economic development and environmental protection require governments to balance financial considerations against social needs, public expectations and political commitments.
An algorithm may calculate which policy produces the greatest measurable financial benefit, but it cannot legitimately determine how much importance a society should attach to fairness, accessibility, environmental protection or the needs of vulnerable groups.
Those judgments remain the responsibility of elected officials who can be questioned, challenged and ultimately held accountable by voters.
Technology companies could gain greater influence
Another concern involves the organizations developing and controlling AI systems.
As government institutions become increasingly dependent on these technologies, the researchers warn that influence over political decision-making could gradually shift toward private companies responsible for designing, maintaining and operating the systems.
Because AI models are largely trained on existing information, they may also reproduce established assumptions and viewpoints, potentially limiting consideration of alternative political ideas.
This creates a challenge for democratic institutions, particularly when the reasoning behind AI-generated recommendations is difficult to examine or explain.
The researchers also identify public trust as a central concern.
When government communications, parliamentary motions and policy proposals are prepared with AI assistance, responsibility for their contents must remain clear.
Citizens must be able to understand how decisions are made and who is answerable when those decisions produce negative consequences.
Boogers and Van de Coevering argue that the future of democracy will depend less on what artificial intelligence is capable of doing and more on the choices governments make about how it is used.
Their central message is that AI may strengthen democratic institutions by improving access to information, but the responsibility for weighing public interests and making political decisions must remain firmly in human hands.