News & Analysis

Gemini Goes Silent on Elections

Google says that it’s AI-led chatbot won’t always answer poll-related queries in the US, India

At a time when GenAI and its larger siblings have polarized the world over both efficacy and subversive possibilities, Google has now announced that its Gemini would restrict answers to queries around elections across the world. Informatively, two of the world’s top democracies – United States and India – go to the polls in 2024. 

According to a report published by TechCrunch, Google has already started the process of setting up restrictions on Gemini across global markets whereby it would limit answers about poll related queries. It seems, the update is already live in the US and the rollout could begin in India very soon and spread to other countries scheduled to have elections. 

Looks like the search giant is trying to win back some goodwill following embarrassments over its large language models (LLM) around image creation. Readers would recall that Google had to pull the plug on Gemini’s image generation component. That too barely a couple of weeks after first going to town with it. 

The latest measure clearly outlines what GenAI has been doing off late and the potential for creating mischief in the future, especially when it comes to the general elections and the power of deep fakes and other misinformation. Proponents of GenAI have been suggesting safeguards while opponents believe that it was only a matter of time before it gets weaponized. 

Of course, one can easily associate Google’s latest move to India’s recent announcement over the need for big tech firms to get sign-offs from government agencies before coming out with new AI models. 

So, what does the new change mean?

For starters, any queries related to political parties, candidates or politicians now return a preset message. You could see something like: “I am still learning how to answer this question. In the meantime, try Google Search”. Of course, smart query engineers found ways to play the tool such as using typing errors while drafting a question. 

In a blog post, Google says, “Out of an abundance of caution on such an important topic, we have begun to roll out restrictions on the types of election-related queries for which Gemini will return responses. We take our responsibility for providing high-quality information for these types of queries seriously, and are continuously working to improve our protections.” 

This update, coming days before India announces its election schedule, was a result of a scandal involving Gemini which described Prime Minister Narendra Modi as a “fascist” accused of implementing such policies. This led to the IT Ministry doing its bit to get such responses removed and then bringing the rule to get government approval for new GenAI models. 

“Protecting the integrity of elections also means keeping our products and services safe from abuse. Across Google, we have long-standing policies to keep our products and platforms safe. Our policies are enforced consistently and apply to all users, regardless of content type,” Google said in the blog post.