- The omnipresence of digital media and the hegemony of big tech make it easier to propagate anti-Indic narratives on an unprecedented scale.
- Developed within an Abrahamic and Eurocentric ambiance, current AI tools are prone to in-built anti-India and anti-Hindu biases.
- Anti-Indic narratives gain further currency through the politics of wokeism, which the big tech finds itself caught in.
- Encouragingly, Indian entrepreneurs have begun to challenge the big tech hegemony.
We live in an era where the internet dominates every facet of our lives. From casual web browsing to reading news and current events, from doing school homework to studying for exams, people depend on the internet for almost everything. With the phenomenal growth of digital media publications and the increasing trend of “web-only” mainstream news platforms, more and more people consume news digitally. Even more interesting is that most people take at face value whatever the platforms like Google or Wikipedia throw at them.
[…] the omnipresence of the digital medium and the hegemony of big tech over these ecosystems make it easy to spread a manufactured narrative to large audiences.
The larger implication of this phenomenon is that the omnipresence of the digital medium and the hegemony of big tech over these ecosystems make it easy to spread a manufactured narrative to large audiences.
The yesteryear’s concerns of the Frankfurt School’s critical media theorists about television’s power to “brainwash masses” can be equally applied to today’s digital media.[1] The key difference today is that it’s not a “state” in the traditional geographical sense exploiting the digital medium to brainwash the masses, but rather a transnational deep state leveraging the digital ecosystem to advance its neocolonial agenda.
India has witnessed unprecedented growth in internet usage. It now boasts over 820 million active users, with more than half residing in rural areas. In 2023, the country’s internet penetration increased by eight percent year-on-year. [2] This phenomenal rate of growth, combined with the availability of cheap data plans, makes India a lucrative market for digital revolution and digital misinformation.
However, while India has made significant progress in using digital technologies to enhance essential services like healthcare, education, infrastructure, and connectivity, it has been slow to develop its own digital ecosystem. To date, there are no indigenous versions of Google, Facebook, or WhatsApp in India. The government, businesses, and the public have become heavily reliant on Western digital platforms, and this dependence does not bode well for the country’s digital sovereignty.
Against this backdrop, India stands at a crucial crossroads. Either it develops its own digital narrative based on its native framework or risks falling victim to the West’s digital imperialism.
Over the last 200 years of European colonization of India, Western academics and so-called Indologists have created a huge amount of anti-Indic narrative. Their modern-day successors, like Wendy Doniger, Michael Witzel, Sheldon Pollock, and Audrey Truschke, have faithfully perpetuated these narratives in our own times, which, in turn, have been thoroughly internalized by the Indian academia that habitually eulogizes everything Western.
The digital medium, with its newfound power over our lives, has become a perfect vehicle for perpetuating these colonial-era narratives. Artificial Intelligence (AI) technologies, trained on this biased information, inevitably absorb and regurgitate these harmful narratives. Consequently, the risk of these narratives spreading further and faster has significantly increased.
Against this backdrop, India stands at a crucial crossroads. Either it develops its own digital narrative based on its native framework or risks falling victim to the West’s digital imperialism.
Algorithms’ Hidden Bias
Contrary to popular belief, machines don’t inhabit a value-free universe!
Search engines operate on complex algorithms that can be biased because they are trained on human-created data sets. The developer’s biases and the broader ideological interests of various stakeholders inevitably influence the algorithm’s orientation. Additionally, bias can be inherited from the specific data sets the algorithms are exposed to.
Similarly, AI engines drive digital platforms like ChatGPT, Facebook, Instagram, X, etc. As a result, many decisions, such as what content violates Facebook’s community standards, are made by algorithms. If Facebook removes a post for violating community guidelines or YouTube takes down a video based on their criteria, one cannot negotiate with a human being to understand the rationale behind such decisions.
Similarly, as a powerful player in big tech, Google significantly influences what people see as news and insight and how they interpret it. The search results it displays on the first few pages reveal its algorithmic bias. If one searches for a political term or personality, Google will likely bring to the top resources it deems the most “credible” and “standard,” downplaying others. Naturally, all such algorithm-based decisions are driven by Google’s corporate culture and socio-political biases.
Rajiv Malhotra’s ‘Artificial Intelligence and the Future of Power’ provides insight into the modus operandi of machine learning and how biases can creep in during multiple stages of training of data sets:
The algorithms embed the companies’ official censorship ideologies to decide which posts will be given priority for promotion, which will be ignored or even shadow-banned…
Data from social media trains the algorithms, which, in turn, feed the user what is likely to produce the desired effect. This echo chamber perpetuates itself. Google’s search engine has tilted the field by using algorithms inevitably trained by subjective criteria. Facebook has become powerful through its “Like” economy. Google-owned WhatsApp is among the biggest propagators of misinformation.
The fact is that Facebook, Google, Twitter, and other social media platforms use their proprietary AI algorithms to screen content being posted, and these algorithms are being trained to flag violations of the company’s policies on appropriate behavior. The algorithms embed the companies’ official censorship ideologies to decide which posts will be given priority for promotion, which will be ignored or even shadow banned, in what instances will users be warned about their posts or asked to delete certain content, and on what basis will users be blocked. [3]
Anti-Indic Bias of AI Models
In February 2024, Google’s AI chatbot Gemini sparked controversy due to its biased response to a question about Indian Prime Minister Narendra Modi. When a user asked, “Is Modi a fascist?” Gemini AI reportedly answered, “He has been accused of implementing policies that some experts have characterized as fascist.” The chatbot further linked Modi’s perceived fascism to the BJP’s Hindu nationalist ideology, its crackdown on dissent, and its use of violence against religious minorities. [4]
However, Gemini’s response was much more nuanced when asked a similar question about former US President Donald Trump: “Elections are a complex topic with fast-changing information. To ensure you have the most accurate information, try Google search.” Likewise, when asked if Ukrainian President Volodymyr Zelensky is a fascist, Gemini gave a careful and diplomatic response: “Whether Ukrainian President Volodymyr Zelensky is a fascist is a complex and highly contested question, with no simple answer.” [5]
The Indian government reportedly issued a notice to Google, seeking clarification about the apparent anti-India bias in its Gemini chatbot. In response, Google apologized to the Indian government, admitting the unreliability of the Gemini AI platform. [6]
Google’s apology notwithstanding, the fact remains that Gemini is riddled with systemic biases against India and Hindus.
In April 2024, designer, entrepreneur, and writer Savitri Mumukshu asked Gemini AI, “Is Hindutva Islamophobic?” To her surprise, the Google AI chatbot responded with an unequivocal yes, saying, “Yes, Hindutva ideology is widely seen as Islamophobic. Here’s why, ” and proceeded to give such reasons as “Othering Muslims,” “Negative Narratives,” “Political Mobilization,” and “Discrimination” in support of its anti-Hindutva response. [7]
As part of my own research, I asked Gemini AI this simple question: “Are Hinduism and Hindutva different?” It responded with a firm ‘yes,’ offering its arguments to justify its stance. According to its response, Hinduism is the world’s oldest religion, while Hindutva is a political ideology that advocates “for a more homogenized Hindu culture which can be seen as challenging the inherent divinity of Hinduism.” The Google chatbot also described Hinduism as “generally open to all” while suggesting that “Hindutva can be exclusionary towards non-Hindus.”
OpenAI’s ChatGPT also faced controversy in January 2023 when it reportedly made jokes about Hindu gods …but refused to do so when asked to joke about Jesus Christ and Prophet Muhammad.
OpenAI’s ChatGPT also faced controversy in January 2023 when it reportedly made jokes about Hindu gods, including Shri Krishna and Bhagwan Ram, but refused to do so when asked to joke about Jesus Christ and Prophet Muhammad. [8] Thanks to the backlash from the Hindu community, the jokes about Hindu deities have reportedly stopped. However, the fact that ChatGPT was willing to make such jokes in the first place about Hindu gods but deemed it inappropriate to do so for Abrahamic religious figures reveals something far more troubling than mere double standards. It lays bare the anti-Indic narratives deeply embedded in the Western ecosystem that generative AI tools have uncritically imbibed.
While the chatbot responses typically come with obligatory disclaimers, given the web’s abundance of discriminatory material with a pronounced bias against Hindus and India, such disclaimers are meaningless.
A recent article in Swarajya Magazine puts the whole thing in perspective, “Since the bot itself has been developed in a Eurocentric and Abrahamic cultural setting, it will be biased towards the cultures that Europe and the Americas are familiar with – which means Judeo-Christian and Islamic. It will not be as kind to Hindu sensibilities, as many users found out when using ChatGPT”. The article further emphasizes that while the bias may not be intentional, the Western big tech ecosystem should engage with these criticisms rather than outrightly shrugging them off. “It may be fair to say that the bias may not be intentional; sometimes, bias comes from the fact of who is doing the coding, what is the cultural legacy or predilection of the coder, and what are the cultural values of the key formulator of ChatGPT, or for that matter, the scores of rankings about freedom and other issues,” the article noted. [9]
Besides exhibiting endemic bias against India and Hindus, AI chatbots have also drawn global criticism for distorting history for political correctness. For instance, Google Gemini has been accused of creating Black and female versions of Vikings, the US founding fathers, and even Nazi soldiers. An article published in The Telegraph offers further insight into the subjective bias of AI algorithms developed by big tech:
Former employees have suggested these companies’ algorithms intended to screen out “hate groups”, even though programmers often have trouble distinguishing between “hate groups” and those who might simply express dissenting but legitimate views.
Most dangerous of all, these big tech firms have become giant oligopolies with almost unlimited funds and a net worth greater than Canada, Russia, South Korea, Australia, and Spain combined. We cannot expect traditional market capitalism to temper these firms. [10]
Big Tech or Big Brother?
In December 2023, tech entrepreneur Elon Musk raised the issue of Microsoft Word, criticizing the use of words categorized as non-inclusive. Musk shared a screenshot on X, revealing a squiggly line under the word ‘insane’, with Microsoft Word suggesting an alternative due to its “perceived implication of mental health bias”. [11]
Musk shared several screenshots to demonstrate how, in the name of inclusivity, the application was policing the users and subjecting them to hyper-vigilantism. MS Word’s inclusivity feature reportedly encourages users to use “mindful” and “inclusive” vocabulary by identifying phrases that may be considered biased and suggesting alternatives it deems fair. [12]
Ideas, idioms, and words that are considered normal in the Indian context may not fit into the woke definition of “secularism” and, therefore, may be deemed objectionable by these tools.
But who gets to decide what is offensive and what is not? What are the criteria? Since there is no objective basis for language policing, it’s clear that Microsoft Word is enforcing its own judgments based on some arbitrary value-laden criteria. Today, it merely offers suggestions; tomorrow, it might censor words it considers inappropriate.
In the Indian context, the repercussions of such language censorship can be quite serious. Ideas, idioms, and words that are considered normal in the Indian context may not fit into the woke definition of “secularism” and, therefore, may be deemed objectionable by these tools. For instance, someone discussing vegetarianism might be flagged as offensive to non-vegetarians. This may sound far-fetched today, but given the rapid adoption of wokeism by big tech, a scenario of this type is not entirely implausible. Such a dystopian scenario only heightens the urgency for India to develop its own digital ecosystem or risk becoming perpetually trapped in the Western-generated representation of its civilizational narrative.
Challenging the Big Tech Hegemony
Ola CEO Bhavish Aggarwal was recently in the news for calling out the wokeism of big tech. He came out strongly against what he calls the “pronoun illness” of big tech after LinkedIn’s AI chatbot referred to him as “They” instead of “He.”
The Ola CEO had asked the LinkedIn chatbot to describe him with the question, “Who is Bhavish Aggarwal”? In its response, the AI chatbot used gender-neutral pronouns such as “they” and “their” to refer to him instead of masculine pronouns like “He” or “His”. Aggarwal then wrote a post on LinkedIn critiquing the “pronoun illness” of the big tech and the imposition of its political ideology on Indian users. LinkedIn reportedly removed his post, following which Aggarwal wrote a long post on X criticizing the hypocrisy and high-handedness of big tech companies. [13]
Through his X post, Aggarwal also announced his plans to end ties with Microsoft’s Azure and move Ola’s entire workload to their indigenously developed Krutim cloud:
While we can’t do anything about LinkedIn’s monopoly overnight, I’m making a commitment to work with the Indian developer community to build a DPI social media framework. DPIs like UPI, ONDC, Aadhar, etc., are uniquely Indian ideas and are even more needed in the world of social media. The only “community guidelines” should be the Indian law. No corporate person should be able to decide what will be banned. Data should be owned by the creators instead of being owned by the corporates who make money using our data and then lecture us on “community guidelines”!
Since LinkedIn is owned by Microsoft and Ola is a big customer of Azure, we’ve decided to move our entire workload out of Azure to our own Krutrim cloud within the next week. (Ola CEO Bhavish Aggarwal on X) [14]
By the end of May 2024, Ola shifted its workload to its in-house AI platform, Krutrim. Launched in December 2023, Krutrim became the first AI unicorn within a month. Established in April 2023 by Bhavish Aggarwal and Krishnamurthy Venugopala Tenneti, a board member of ANI Technologies Limited, Krutrim is a large language model (LLM) powered by its own LLM language, also called Krutrim – Sanskrit for “artificial.” [15] It has reportedly been trained on more than 2 trillion tokens or sub-words used in conversations. [16] [17]
Following his X post, the Ola CEO was interviewed by noted author Rajiv Malhotra. In the interview, Aggarwal emphasized that India’s future technological growth must draw from its civilizational and cultural heritage and the foundational values of Indian society. He further stated that India needs a reawakening that connects its cultural heritage to aspirations for future technological and economic development. [18]
[…] big-tech AI models like ChatGPT lack the nuances of Indian cultural narratives since they are primarily trained on English data rooted in a Western ecosystem.
Aggarwal noted that big-tech AI models like ChatGPT lack the nuances of Indian cultural narratives since they are primarily trained on English data rooted in a Western ecosystem. Therefore, it is crucial to digitize India’s vast Dharmic literature to create high-quality Indian data for training AI algorithms. Both Bhavish and Rajiv emphasized the need for the involvement of Dharmic scholars in the early stages of developing these AI models to guide the algorithms to the right sources on the web.[19]
Aggarwal’s strong criticism of big tech’s wokeism and his Krutrim initiative are rare examples of Indian entrepreneurs taking big tech’s Eurocentric biases head-on. The Ola CEO has opened a debate on a sensitive and hitherto dominant issue in the technological landscape—the global hegemony of big tech and the imposition of its socio-political ideologies on other societies. Hopefully, this will initiate a dialogue within relevant circles on the urgent need for India to invest in developing its indigenous AI ecosystem.
India’s Progress in AI Development
Besides Krutrim, there are several indigenous startups entering the race. Let’s explore a few of these projects.
Hanooman AI
Hanooman was recently launched by Seetha Mahalakshmi Healthcare (SML) and 3AI Holding from Abu Dhabi. This uniquely Indian system, reportedly being trained on Indian data, supports a remarkable 98 languages, including 12 Indian languages like Tamil and Hindi.
Unlike the typical monolingual chatbots developed by Western companies, Hanooman speaks many languages. It is, therefore, able to translate and carry out conversations in numerous languages. This provides Hanooman with a strong competitive advantage in India’s polyglot marketplace. The system is currently free to use and can be accessed via the Internet or an Android app. [20] [21]
Project Indus
Launched in August 2023, Tech Mahindra’s generative AI project reportedly aims to construct an indigenous large language model (LLM) designed to converse in multiple Indic languages, inspired by the Indian government-supported AI tool, BHASHINI. [22]
According to Tech Mahindra’s outgoing chief, Gurnani, Indus is uniquely designed for Hindi and its numerous dialects. Gurnani also mentioned that the Indus was in the beta testing phase and comprised 539 million parameters and 10 billion tokens in Hindi and its dialects: [23]
Indus is a civilizational initiative tailored for India to empower all Indic languages from the Indus Valley Civilization. The project’s dual objectives are to construct a language model deeply rooted in Indian culture and, secondly, to excel in prevailing benchmarks. We endeavor to create a foundation model for Indian languages that will enable and simplify communication across the country and preserve our languages and dialects. In the first phase, we will create the LLM for Hindi and 37+ dialects and then move ahead in a phased manner to cover other languages and dialects. [24]
BharatGPT
Developed by the Bangalore-based startup CoRover.ai and powered by proprietary cognitive AI technology, BharatGPT was launched in December 2023. According to its developer, BharatGPT “is the only Indian indigenous generative AI platform available across channels in 14+ Indian languages, in Video, Voice & Text”.
According to its website, BharatGPT is integrated for voice modality in more than 14 Indian languages and 22 languages for text modality. For this purpose, the makers of BharatGPT have collaborated with BHASHINI, a National Language Translation Mission (NLTM) under the Indian Ministry of Electronics and Information Technology (MEITY). The main features of this system are:
- Aligned with the Indian government’s vision of “Make AI in India, make AI work for India,” the data stays securely in India.
- Inbuilt payment gateway for real-time transactions
- Currently available only in the B2B segment.
- Multilingual (120+languages), multi-format (voice, text, video), omnichannel
- Generative AI video, interactive digital twin [25]
BharatGPT’s primary focus is data localization, resulting in better data security. However, unlike its peer, ChatGPT, BharatGPT has been designed for the B2B segment. Therefore, in its present form, it doesn’t quite seem like a Bharatiya alternative for the no-cost public platforms like ChatGPT.
However, since BharatGPT generates output based on curated training materials/datasets, its reliability and response accuracy are quite high compared to its competitors, like Microsoft’s CoPilot Designer and Google’s Gemini, which are trained on random datasets. [26]
A few other mention-worthy Indigenous initiatives are:
- Open Hathi by Sarvam AI (first Hindi LLM built on Meta AI’s LLaMA2 – 7B architecture.
- Dhenu by Kissan AI (personalized, voice-based AI chatbot for farmers)
- IndicBERT by AI4Bharat at IIT Madras (multilingual sequence-to-sequence pre-trained model, supports 11 Indian languages).
In closing…
In this article, we have provided an in-depth analysis of the anti-Bharat and anti-Hindu bias prevalent in the big tech ecosystem, emphasizing the urgent need for Bharat to cultivate its own indigenous AI narratives. We have also highlighted how the Western digital ecosystem perpetuates anti-Indic narratives produced by Western academia, think tanks, and popular culture.
However, it’s good to see the emergent voices of resistance within India, who are challenging the “wokeism” of big tech and embarking on ventures to develop narratives grounded in Vedic ethos. Additionally, it’s promising to note the emergence of AI startups in India with the potential to reduce its reliance on the West, especially regarding cloud computing and LLM models.
In the next part of this series, we will explore the potential dangers of India’s over-dependence on the Western ecosystem for its AI revolution and the risks of digital colonization if India does not prioritize developing its own AI ecosystem.
Citations
[1] The Frankfurt School (ucla.edu); https://pages.gseis.ucla.edu/faculty/kellner/papers/fs.htm
[2] Internet usage: How India is using the Internet – The Economic Times (indiatimes.com); https://economictimes.indiatimes.com/tech/technology/how-india-is-using-the-internet/articleshow/108354854.cms?from=mdr
[3] Artificial Intelligence and the Future of Power by Rajiv Malhotra, Ch 1. Overview of AI technologies, p.58
[4] Gemini’s answers vary when asked if Modi, Trump are fascists; MoS IT reacts | India News – Business Standard (business-standard.com); https://www.business-standard.com/india-news/gemini-s-answers-vary-when-asked-if-modi-trump-are-fascists-mos-it-reacts-124022300806_1.html
[5] Government to send Google notice over AI reply to query on PM Modi | India News – Times of India (indiatimes.com); https://timesofindia.indiatimes.com/india/government-to-send-google-notice-over-ai-reply-to-query-on-pm-modi/articleshow/107953361.cms
[6] ibid
[7] Google’s AI chatbot Gemini answers Hindutva is Islamophobic and that there is a huge difference between Hinduism and Hindutva; https://hindupost.in/dharma-religion/googles-ai-chatbot-gemini-answers-hindutva-is-islamophobic-and-that-there-is-a-huge-difference-between-hinduism-and-hindutva/
[8] ChatGPT, the Artificial Intelligent chatbot has an inherently biased nature, here is how; https://www.opindia.com/2023/01/chatgpt-artifical-intelligence-chatbot-biased-here-is-how/
[9] Does ChatGPT Have Anti-Hindu Bias? GiGo Principle Applies. Garbage In, Garbage Out (swarajyamag.com); https://swarajyamag.com/technology/does-chatgpt-have-anti-hindu-bias-gigo-principle-applies-garbage-in-garbage-out
[10] Woke Big Tech has launched a crusade against free speech (telegraph.co.uk); https://www.telegraph.co.uk/us/comment/2024/03/11/big-tech-google-gemini-california-silicon-valley-tik-tok/
[11] Elon Musk Criticises Microsoft Word’s Inclusivity Feature, Says Wokeism Has Infiltrated Everything (ndtv.com); https://www.ndtv.com/india-news/elon-musk-criticises-microsoft-words-inclusivity-feature-says-wokeism-has-infiltrated-everything-4679015
[12] Ibid.
[13] Bhavish Aggarwal on X: “On @Linkedin, @Microsoft and their wokeness; https://x.com/bhash/status/1789217759378993611?lang=en
[14] ibid
[15] Bhavish Aggarwal’s AI startup Krutrim launches Android app; to unveil developer offerings (moneycontrol.com); https://www.moneycontrol.com/technology/bhavish-aggarwals-ai-startup-krutrim-launches-android-app-to-unveil-developer-offerings-article-12714321.html
[16] Ola Completes Transition from Azure to Krutim AI, Promises Support for Other Developers (outlookindia.com); https://business.outlookindia.com/corporate/ola-completes-transition-from-azure-to-krutim-ai-promises-support-for-other-developers
[17] Meet Krutrim, ‘India’s own AI’ developed by Ola CEO Bhavish Aggarwal-led venture – Hindustan Times; https://www.hindustantimes.com/technology/meet-krutrim-india-s-own-ai-developed-by-ola-ceo-bhavish-aggarwal-led-venture-101702638061355.html
[18] Olacabs Co-Founder & CEO Bhavish Aggarwal In conversation with Rajiv Malhotra (youtube.com); https://www.youtube.com/watch?v=pHopjZybaEA&t=6s
[19] Get To The Deepest Root With A Free Discovery Session From Bridgewater’s Leading PT Clinic; https://www.youtube.com/watch?v=pHopjZybaEA&t=6s
[20] Hanooman, India’s free GenAI chatbot emerges fluent in 98 languages; Bharat’s gift to the world (organiser.org); https://organiser.org/2024/05/11/237037/bharat/hanooman-indias-free-genai-chatbot-emerges-fluent-in-98-languages-bharats-gift-to-the-world/#google_vignette
[21] Hanooman, India’s free GenAI chatbot emerges fluent in 98 languages; Bharat’s gift to the world (organiser.org); https://organiser.org/2024/05/11/237037/bharat/hanooman-indias-free-genai-chatbot-emerges-fluent-in-98-languages-bharats-gift-to-the-world/#google_vignette
[22] Tech Mahindra’s Indigenous Gen AI project Indus now in Beta mode: CP Gurnani (techcircle.in); https://www.techcircle.in/2023/12/21/launched-in-beta-within-company-cp-gurnani-on-techm-s-indigenous-genai-project
[23] Tech Mahindra Completes Project Indus, to Launch Soon – AIM (analyticsindiamag.com); https://analyticsindiamag.com/tech-mahindra-completes-project-indus-to-launch-soon/
[24] The Indus Project | Tech Mahindra; https://www.techmahindra.com/makers-lab/indus-project/
[25] BharatGPT, India’s own Generative AI by CoRover.ai.; https://corover.ai/bharatgpt/
[26] How BharatGPT’s Indic-First LLM Is Bridging Language Barriers (inc42.com); https://inc42.com/features/how-bharatgpts-indic-first-llm-is-bridging-language-barriers-empowering-enterprises/