Archive — US — techcrunch
Google still limits how Gemini answers political questions
Google's AI chatbot, Gemini, continues to limit responses on political topics, standing apart from competitors like OpenAI and Anthropic that have relaxed similar restrictions. Despite recent testing revealing Gemini's struggles with factual political information, Google has not announced changes to its policy, raising concerns about censorship and the chatbot's effectiveness.
Key points
- Google's Gemini won't answer political questions, stating it 'can't help with responses on elections and political figures.'
- Competitors such as OpenAI, Anthropic, and Meta AI provide more response options on political queries.
- Google's cautious stance is attributed to concerns about backlash from inaccurate responses.
- Testing found Gemini provided erroneous information about U.S. political figures, such as misidentifying the current president.
- Google acknowledged Gemini's confusion regarding political terms and is working on improvements.
- There is criticism of Google's approach as being overly conservative, especially from figures in Silicon Valley advocating for less censorship.
Why it matters
- AI companies adapted their policies following backlash from previous electoral cycles, leading to self-censorship.
- The political landscape's complexity raises the importance of AI's ability to handle sensitive topics fairly.
- Historical context shows fluctuating media and technology responses to political discourse, influencing current AI models.
- OpenAI and other companies are now promoting 'intellectual freedom' to address censorship accusations.
Google's conservative approach with Gemini highlights the ongoing debate over AI censorship and the capability of AI to handle complex political questions accurately. As competitors move towards more open discourse, Google may need to reassess its strategy to maintain relevancy.