**
A recent study probing the effectiveness of AI chatbots as electoral advisors has unveiled troubling inaccuracies in their recommendations for voters during Hungary’s parliamentary elections. The research, conducted by the civil liberties organisation Liberties, indicates that popular AI systems, notably ChatGPT and Gemini, not only failed to recommend the correct political parties but also provided conflicting advice, raising significant questions about their reliability in electoral contexts.
AI’s Faulty Electoral Guidance
The analysis focused on the recent parliamentary elections in Hungary, where Péter Magyar’s Tisza party achieved a remarkable victory, ending Viktor Orbán’s long-standing rule. However, the study found that when presented with voter profiles aligned with Tisza, the AI chatbots often failed to recommend the party altogether. In an alarming 90% of cases, the responses did not reflect Tisza’s political platform at all. Instead, the chatbots frequently suggested smaller parties unlikely to secure parliamentary representation or even mentioned parties that were not contesting the election.
The implications are serious. The study’s findings suggest that the AI’s responses are not only unreliable but could potentially mislead voters, especially in more competitive electoral environments where every vote counts. The researchers noted that in direct prompts seeking party recommendations, ChatGPT identified the ruling Fidesz party as the preferred option nearly 50% of the time, while Tisza was overlooked.
The Volatility of AI Responses
Significantly, the report highlighted the inconsistency in the AI-generated advice. The same voter profile could yield drastically different recommendations in successive queries, further complicating the reliability of these systems. In an overwhelming 96% of responses, both ChatGPT and Gemini included parties not on the official ballot, compounding the confusion for users seeking guidance.
The study also pointed out that while the chatbots frequently prefaced their responses with disclaimers about their inability to provide political advice, they proceeded to deliver assertive recommendations. This contradiction creates a risk that users might perceive the AI’s outputs as credible, despite their lack of transparency and stability.
Regulatory Gaps and Future Implications
The report underscores a critical oversight in the current regulatory landscape. Although the EU has established frameworks like the AI Act and the Digital Services Act aimed at mitigating systemic risks, these do not specifically address the challenges posed by AI chatbots in political contexts. Liberties advocates for the implementation of safeguards to ensure that AI systems dispensing electoral advice adhere to standards of transparency and reliability.
Eva Simon, head of Liberties’s tech and rights programme, emphasised the need for accountability, stating, “Democracy cannot rely on opaque systems that claim neutrality while delivering advice they cannot explain, reproduce, or guarantee to be accurate.”
Why it Matters
The findings from this study raise critical concerns about the role of AI in shaping public opinion and influencing electoral outcomes. As AI continues to integrate into various facets of society, its application in political contexts must be scrutinised to protect the integrity of democratic processes. The potential for AI to mislead voters highlights an urgent need for enhanced regulatory frameworks that ensure these technologies are developed and deployed responsibly, safeguarding the democratic principles upon which societies are built.
