A recent study examining AI chatbots' performance during Hungary's parliamentary elections revealed that systems like ChatGPT and Gemini provided inaccurate, unreliable, and inconsistent voting advice. The research, conducted by the civil liberties group Liberties, found that these AI models frequently failed to recommend the correct parties, sometimes suggested parties not even participating in the election, and offered vastly different answers to identical prompts. While the study does not claim the AI's output affected the election's outcome, it warns that such unreliability could influence more competitive races and highlights a significant regulatory gap concerning the use of general-purpose AI in electoral contexts. AI
IMPACT Highlights the unreliability of current general-purpose AI for sensitive tasks like political guidance, potentially impacting voter trust and electoral integrity.
RANK_REASON Research paper detailing inaccuracies in AI chatbot performance on a specific task (election advice).
Read on Mastodon — fosstodon.org →
- Ai Chatbots
- Frontier Ai
- Hungary
- Mercor
- The Guardian
- ChatGPT
- Fidesz party
- Gemini
- Liberties
- Mastodon
- Péter Magyar
- Viktor Orbán
- Voksmonitor
AI-generated summary · Google Gemini · from 6 sources. How we write summaries →
