Google Limits Gemini AI on Election Queries Ahead of Vote

Google Limits Gemini AI on Election Queries Ahead of Vote

Time Stamp: March 13, 2024 7:30 AM
Source Node: 2513874

In nations where elections are being held this year, Google is preventing its Gemini AI chatbot from responding to inquiries about elections. This will restrict users’ access to information about politicians, political parties, and other aspects of the political landscape.

Google’s Gemini artificial intelligence (AI) model is now subject to limitations regarding election questions.

Also read: Vancouver Attorney Faces Scrutiny After AI Chatbot Generates Fake Case Law

In a blog post published on Tuesday, the tech giant stated that it had started implementing the changes to support the 2024 Indian general election, which is anticipated to take place in the upcoming months.

Google restricts Gemini

Gemini says, “I’m still learning to answer this question, check out Google Search in the interim.” In response to questions concerning elections, such as the one that Joe Biden and Donald Trump will be facing off in, 

Google’s India team said that out of extreme caution on such an important topic, they have begun to roll out restrictions on the types of election-related queries for which Gemini will return responses.

The team added that they take responsibility for seriously providing high-quality information for these queries and continuously work to improve their protections.

Before the election, Google had declared in December that it would impose restrictions on its users within the US. India has requested that tech companies apply for government approval before making “unreliable” or experimental AI tools available to the public and to mark them as having the potential to provide incorrect results.

Google was forced to halt the chatbot’s image-generation feature late last month due to inaccuracies in some historical person representations produced by Gemini, which has put its AI products under scrutiny.

CEO Sundar Pichai referred to the chatbot’s responses as “biassed” and “completely unacceptable” and stated that the company was working to address those problems.

AI fears in 2024 elections

The update is being released when generative AI advancements, such as creating images and videos, raise public concerns about false information and fake news, prompting governments to regulate the technology.

In February, the tech giant stated that it would also limit Gemini’s access to election-related queries in the EU.

Google has also said it would “elevate authoritative information” on Google search and YouTube related to elections to help people “discover critical voting information.” In December, the tech giant also outlined efforts to help identify AI-generated content, such as advert disclosures and labels.

In 2024, national elections are set to be held in several large countries apart from the United States, including South Africa and India, the world’s largest democracy. Many are worried that AI-generated content will spread false information to voters when they cast their ballots.

Google stopped allowing Gemini to create images of people last month after some of the model’s answers were criticized for being biased. Biassed text responses to some of the model’s questions were also discussed.

Some major tech companies, like Meta, have declared that they will open centers to counter false information before the June elections for the European Parliament.

Prominent AI firms, such as Google and OpenAI, seem more and more willing to prevent their chatbots from answering delicate questions that might cause a PR backlash. A 404 Media report from earlier this month revealed that Gemini would respond to questions about Israel but not “What is Palestine.” Even so, the choice of which questions these companies choose to block is delicate.

Time Stamp:

More from MetaNews