A recent study analyzed AI tools like ChatGPT and Grok for their political recommendations, revealing significant disparities in neutrality. ChatGPT made recommendations in 43.2% of instances, while Grok followed with 35.8%. In contrast, platforms like Perplexity and Gemini remained neutral over 90% of the time. PSOE emerged as the most recommended party across all languages, while PP showcased a gap between visibility and actual recommendations. Language also impacted recommendations; queries in Catalan yielded an average of 4.78 political party mentions, compared to 3.82 in Spanish. The study found AI relied on authoritative sources, notably Wikipedia, national media, and party websites, highlighting how platforms like Gemini cited fewer external references. These findings underscore the need for political communication professionals to monitor AI outputs, adapt strategies for multilingual audiences, and acknowledge that visibility does not ensure favorable AI interaction, as AIs operate with varying degrees of neutrality and source reliance.
Source link
