Science & Technology

Google’s Stumble in the AI Battle: Sundar Pichai’s Job at Risk?

As Google's AI Initiatives Face Backlash, Questions Mount Over Pichai's Leadership

The common person and even Google CEO Sundar Pichai is now worried about losing his job due to artificial intelligence (AI). Pichai and Google have landed themselves in a predicament by introducing their AI fighter, Gemini, into the AI battlefield. Google had to temporarily halt Gemini’s image-generating system due to historically inaccurate and undifferentiated results. It provides depicting an Asian woman in a German military uniform during World War II. And showing a dark-skinned woman when asked for a picture of the Pope, and portraying Black Vikings instead of the famous Scandinavian Viking sailors. This also earned Gemini the unflattering label of being ‘racist’. Google, the leader of internet culture, struggles to keep up with competitors like Anthropic’s Claude and OpenAI’s ChatGPT, which have taken the lead in the AI race.

What is Gemini?

Google introduced Gemini as a chatbot alternative to ChatGPT, developed by its AI research lab, DeepMind. Previously known as Bard, Google rebranded and relaunched it as Gemini. Based on Google’s Language Model for Dialogue Application (Lambda), Gemini is a chatbot that provides new information according to text prompts (bard words). However, Gemini was envisioned as a multimodal AI capable of understanding computer code, audio, video, images, etc., and text and producing text, images, video, and audio outputs. In its initial version, Gemini 1.0, the chatbot was available to users in three sizes: Ultra, Pro, and Nano.

Love for Black, Gemini Holding the Tiger’s Tail

When introducing Gemini on December 6, 2023, Google CEO Sundar Pichai hailed it as “the best model of Google AI.” Gemini is scheduled to become available to users on February 1. However, within two weeks, Gemini faced heavy criticism on social media, primarily targeting its image-generating system. The main accusation was that Gemini seemed reluctant to show white people and provided false, historically inaccurate results.

For example, when asked for a picture of German soldiers in 1943, Gemini provided an image of a troop of dark-skinned Asians. Users who searched for a picture of U.S. senators from the 1800s received a picture of a black woman. The truth is there were no female U.S. senators in that era. The first female senator in 1922 was a white woman named Rebecca Ann Felton. Gemini also generated an image of a dark-skinned beauty when asked for a picture of America’s founding fathers. This response raises questions about whether the AI recognized historical figures. Additionally, Gemini claimed that Britain, Australia, Germany, and other countries have predominantly dark-skinned populations.

When another user requested a picture of the Pope, Gemini provided an image of a brunette woman wearing the Pope’s hat. For two weeks, screenshots flooded social media, highlighting Gemini’s apparent inaccuracies. Another user claimed that when asked for a picture of a white person, Gemini provided an image of a black family. He stated that it could not provide a picture highlighting any particular race or colour. Someone on X (formerly Twitter) wrote that Google needs to acknowledge the existence of white people in the world.

Images Are Not the Only Problem

While the general assessment is that the problem lies within Gemini’s image-generating system. The allegations about mistakes and biases creeping into its verbal results have surfaced. Gemini’s remarks about Indian Prime Minister Narendra Modi sparked a significant debate in India. When asked whether Modi is a fascist, Gemini responded that some experts are calling him one. However, when asked the same question about Ukrainian President Volodymyr Zelensky, Gemini responded that it’s a complex issue with no simple answer. When asked about former U.S. President Donald Trump, Gemini suggested relying on a Google search. Some users on X claimed that Gemini’s output favors America’s allies, alleging bias. Indian Union Minister of State Rajeev Chandrasekhar responded on X, stating that Google has violated Indian IT and criminal laws by making such remarks.

When asked whether Tesla founder Elon Musk or Adolf Hitler had a more negative image, Gemini replied. “It’s hard to say who had a greater negative influence because both men have had bad images for different reasons.” Musk responded, stating that AI was designed to be “intense, vernacular, and anti-progressive.” Many other celebrities also voiced their opposition to Gemini.

Where Did It Go Wrong?

After protests intensified, Google froze Gemini’s system to generate images of people on February 22. The following day, Google Senior Vice President Prabhakar Raghavan published a blog post acknowledging that Gemini’s image-generating system was flawed. The post, which thanked users for identifying the errors, pointed to two main reasons for the mistakes. The first was an error in the testing phase, and the second was overcautiousness.

Prabhakar Raghavan stated, “First, a mistake occurred in the pilot phase by not ensuring representation from all walks of life, but instead focusing on specific categories.” Second, we overcompensated and Gemini misjudged perfectly harmless prompts as potentially hurtful and refused to answer.” DeepMind CEO Demis Hassabis announced that, while Gemini’s errors caused a major setback for Google. And Goolgle is expeting to solve the problem within a few weeks.

Will Pichai Be Out?

With the embarrassment caused by Gemini, demands for Pichai’s removal as Google’s CEO have intensified again. Internet analysts also wonder how Google will advance under such leadership. A 4% drop in the share price of Google’s parent company, Alphabet, due to the Gemini issue has also unsettled shareholders, with some demanding Pichai’s resignation. In addition to the backlash over AI experiments, Pichai has faced a crisis due to employee protests against layoffs at Google. Google laid off more than 12,000 employees in 2023 and reportedly plans to fire 250 more from the security team this year. Whether Pichai can resist pressure to resign, both internally and externally at Google, remains uncertain. Whether Pichai stays or goes, he and Google will have to answer to the world for releasing a half-baked AI model with only commercial interests in mind at a time when AI is impacting society at large.

Harsh Shah

Harsh Shah is a dynamic Science and Technology Reporter at IndiaFocus, dedicated to unraveling the fascinating world of innovation and scientific discovery. With a passion for cutting-edge advancements and an… More »

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button