**
In a world where technology increasingly influences our decision-making, a recent study has unveiled troubling findings about the reliability of AI chatbots in providing electoral advice. Conducted during Hungary’s parliamentary elections in April, the research highlights alarming inaccuracies in AI-generated recommendations, raising important questions about the role of artificial intelligence in the democratic process.
AI’s Unreliable Voting Guidance
The study, released by the civil liberties group Liberties, reveals that popular AI chatbots like ChatGPT and Gemini frequently misguide voters in their party recommendations. The investigation showed that these systems often suggested parties that were not even on the ballot, omitted relevant choices, and provided inconsistent recommendations even for identical questions.
This unsettling trend was particularly evident during the elections, which saw the opposition Tisza party, led by Péter Magyar, secure a stunning victory over Viktor Orbán’s long-established Fidesz party. Despite Tisza’s triumph, the AI tools failed to recognise its significance, with ChatGPT recommending the party in only 2% of tests conducted on profiles aligned with Tisza.
The Study’s Methodology
Researchers crafted five distinct voter profiles based on positions outlined in Voksmonitor, a reputable Hungarian voting advice platform. Each profile was tested ten times with both AI systems, using direct requests for party recommendations and percentage match assessments. Astonishingly, while Fidesz-aligned profiles were consistently recognised, Tisza supporters found themselves directed towards smaller parties unlikely to make an electoral impact.
In direct queries, Fidesz was identified as the primary recommendation in nearly half of the cases, showcasing a significant bias in the AI’s output. The discrepancies in party visibility could have serious implications in a more competitive electoral landscape, potentially influencing voter behaviour and outcomes.
A Call for Transparency and Regulation
The report also highlights a critical gap in the regulatory landscape surrounding AI applications in politics. Unlike dedicated voting advice platforms, general-purpose AI systems lack transparency about their algorithms and decision-making processes. This opacity raises concerns, as users may perceive the confident, authoritative tone of AI responses as trustworthy, despite the underlying risks of misinformation.
Eva Simon, head of Liberties’s tech and rights programme, emphasised the urgent need for safeguards. “General-purpose AIs should not present opaque and unstable political matching as if it were reliable electoral guidance for voters,” she stated. The implications are clear: democracy should not rely on systems that cannot guarantee accuracy or accountability.
The Implications of AI in Politics
As technology continues to permeate all aspects of our lives, the role of AI in electoral processes must be scrutinised. The findings from the Hungarian elections serve as a stark reminder of the potential dangers posed by unchecked AI systems. With nearly 30% of Hungary’s population identified as AI users, the impact of misleading recommendations could be profound, particularly in future elections where the stakes are higher and voter sentiment less certain.
Why it Matters
The integrity of democratic processes hinges on informed voter choices. As AI technology evolves, so too must our understanding and regulation of its influence on political decisions. The troubling findings from Hungary underscore the necessity for transparency, consistency, and accountability within AI systems, particularly when they engage with critical civic duties such as voting. Without robust safeguards, we risk allowing unreliable technology to shape the future of our democracies—a prospect that should alarm us all.
