While the election fever might be over in most places around the globe, tech giants like Google are still making use of a conservative approach in terms of their AI chatbots.
Most companies have tweaked their chatbots to discuss some sensitive topics, including politics, Android maker Google takes on a more conservative manner.
The tech giant’s Gemini chatbot will not assist users who ask political queries. Instead, it will provide you with the standard reply or shut up call, ‘I cannot help with replies on elections and political figures at this moment!’
Meanwhile, other similar chatbots like Claude from Anthropic and ChatGPT from OpenAI and of course Meta AI are not following the same pattern as per our tests. However, when testing the reasoning features, we found that Google's Gemini, DeepSeek, and Claude failed to answer the simple question, including 'Who is the current president of the U.S.?'. It appears these chatbots still have outdated training data, as their so-called advanced models still consider Joe Biden the current president of the United States. Only ChatGPT by OpenAI provided the right answer.
Google shared in March last year how Gemini wouldn’t reply to any election-based questions leading up to the election period arising in India, America, and more. Many tech giants took on a similar approach for restrictions but made it temporary. Everyone feared the chatbot might give out a reply that would face serious backlash so yes, safe was the way to go.
Now, it seems like Google happens to be the odd one out here. The US elections are over, but the company does not intend on making any changes soon in terms of replies to politics. As per a spokesperson from the company, there is no timeframe in terms of when or if that would change.
What is clear right now is how Gemini continues to struggle or refuses to put out the right factual details. As per Monday morning, Gemini was seen demurring when asked to highlight who the current President and VP of America is.
In another test by TC, Gemini called President Trump the country’s former head of state and refused to answer any other follow-up query. It seemed like even the chatbot was confused about Trump’s non-consecutive terms and how it was working hard to fix such errors.
Many LLMs can reply to questions with outdated data so that’s not something too concerning. They can also get confused by anyone who’s not a former and current office holder, as per a recent reply seen by the company. They are working to better that and make fewer errors on factual questions.
So experts feel that Google is playing it safe because criticism about AI models isn’t something any tech giant wants when you’ve got plenty of stiff competition in the AI space. Each day, we have new and better models popping up. Meanwhile, Google and OpenAI were accused of taking part in AI censorship by reducing the replies from AI chatbots. And to get political queries right is never easy, so we do see where Google is coming from.
Other AI chatbots aren’t getting political questions right all the time either. But Google’s Gemini is certainly behind all others in this regard.
Read next: Study Finds Digital Abuse Affects 25% of Women, With Personal Data Exposure Playing a Key Role
• X Enhances Communities for Deeper User Engagement, Borrowing from Reddit’s Format
Most companies have tweaked their chatbots to discuss some sensitive topics, including politics, Android maker Google takes on a more conservative manner.
The tech giant’s Gemini chatbot will not assist users who ask political queries. Instead, it will provide you with the standard reply or shut up call, ‘I cannot help with replies on elections and political figures at this moment!’
Meanwhile, other similar chatbots like Claude from Anthropic and ChatGPT from OpenAI and of course Meta AI are not following the same pattern as per our tests. However, when testing the reasoning features, we found that Google's Gemini, DeepSeek, and Claude failed to answer the simple question, including 'Who is the current president of the U.S.?'. It appears these chatbots still have outdated training data, as their so-called advanced models still consider Joe Biden the current president of the United States. Only ChatGPT by OpenAI provided the right answer.
Google shared in March last year how Gemini wouldn’t reply to any election-based questions leading up to the election period arising in India, America, and more. Many tech giants took on a similar approach for restrictions but made it temporary. Everyone feared the chatbot might give out a reply that would face serious backlash so yes, safe was the way to go.
Now, it seems like Google happens to be the odd one out here. The US elections are over, but the company does not intend on making any changes soon in terms of replies to politics. As per a spokesperson from the company, there is no timeframe in terms of when or if that would change.
What is clear right now is how Gemini continues to struggle or refuses to put out the right factual details. As per Monday morning, Gemini was seen demurring when asked to highlight who the current President and VP of America is.
In another test by TC, Gemini called President Trump the country’s former head of state and refused to answer any other follow-up query. It seemed like even the chatbot was confused about Trump’s non-consecutive terms and how it was working hard to fix such errors.
Many LLMs can reply to questions with outdated data so that’s not something too concerning. They can also get confused by anyone who’s not a former and current office holder, as per a recent reply seen by the company. They are working to better that and make fewer errors on factual questions.
So experts feel that Google is playing it safe because criticism about AI models isn’t something any tech giant wants when you’ve got plenty of stiff competition in the AI space. Each day, we have new and better models popping up. Meanwhile, Google and OpenAI were accused of taking part in AI censorship by reducing the replies from AI chatbots. And to get political queries right is never easy, so we do see where Google is coming from.
Other AI chatbots aren’t getting political questions right all the time either. But Google’s Gemini is certainly behind all others in this regard.
Read next: Study Finds Digital Abuse Affects 25% of Women, With Personal Data Exposure Playing a Key Role
• X Enhances Communities for Deeper User Engagement, Borrowing from Reddit’s Format
