Imagine a world where the wisdom of centuries, rooted in diverse cultures and traditions, is slowly fading away, overshadowed by the dominance of a single, homogenized perspective. This is not a distant dystopian future but a reality we are already facing, thanks to the rapid rise of Artificial Intelligence (AI) and its impact on global knowledge systems. But here's where it gets controversial: while AI promises to revolutionize how we access information, it may also be inadvertently erasing invaluable knowledge, particularly from marginalized communities and non-Western cultures. And this is the part most people miss: the very systems designed to make knowledge more accessible could be contributing to a global 'knowledge collapse.'
A few years ago, my family faced a critical decision when my father was diagnosed with a tumor on his tongue. My older sister, a Western-trained doctor, strongly recommended surgery, while my parents, rooted in traditional Indian remedies, preferred herbal treatments. As the family mediator, I turned to the internet for guidance, ultimately siding with my sister. However, my father secretly continued his herbal regimen, and to our surprise, the tumor shrank and disappeared. This personal experience made me question: Are we too quick to dismiss traditional knowledge in favor of digitally dominant sources?
This dilemma is not just personal; it reflects a broader issue in the age of AI. At Cornell University, my research on responsible AI systems has revealed how digital platforms and generative AI (GenAI) amplify existing power imbalances in knowledge. The early internet, dominated by English and Western institutions, has left vast amounts of human knowledge undigitized. Now, GenAI, trained on this biased corpus, risks entrenching these disparities further.
Here’s the controversial part: GenAI, while appearing neutral, often privileges Western and institutional ways of knowing, marginalizing oral traditions, embodied practices, and 'low-resource' languages like Hindi or Swahili. A 2025 study on ChatGPT usage showed that while half the queries sought practical information, the responses were skewed toward dominant knowledge systems. This isn’t just about representation; it’s about the resilience and diversity of knowledge itself.
Languages, as vessels of knowledge, carry centuries of human experience. Yet, 97% of the world’s languages are classified as 'low-resource' in computing, despite having millions of speakers and rich traditions. For instance, Tamil, my mother tongue, spoken by over 86 million people, represents just 0.04% of Common Crawl’s data. This digital exclusion threatens to erase unique ways of understanding the world, from local ecologies to Indigenous architectural knowledge.
Consider Dharan Ashok, an architect reviving natural building techniques in India. He highlights how traditional methods, rooted in local plants and oral traditions, are being lost due to a lack of documentation and the dominance of modern, Western-centric construction practices. Similarly, Bengaluru’s water crisis exemplifies how community-led water management systems, once effective, were replaced by centralized, Western-inspired models, leading to ecological and social disruptions.
But here’s the thought-provoking question: Are we willing to challenge the cultural hegemony that prioritizes Western knowledge systems? As Linda Tuhiwai Smith argues in Decolonizing Methodologies, colonialism disrupted local knowledge systems by severing ties to land, language, and history. Today’s AI systems, built on this distorted foundation, risk perpetuating these inequalities.
AI models, trained on massive datasets, inherently amplify dominant ideas while sidelining niche or marginalized knowledge. This ‘mode amplification’ creates a feedback loop where underrepresented knowledge becomes even less visible. For example, an AI chatbot serving farmers in Asia and Africa relies on government and institutional databases, excluding effective local practices that lack formal validation. This isn’t just a technical issue; it’s a structural one, shaped by liability concerns and power dynamics.
The consequences are far-reaching. As AI integrates into education, like Khanmigo in Karnataka, future generations may lose touch with local wisdom essential for sustainable living. The climate crisis, already exposing cracks in dominant paradigms, demands that we reconsider the value of diverse knowledge systems. Yet, AI developers often prioritize commercial interests and Western cultural values, further marginalizing alternative perspectives.
So, what’s the solution? Incorporating more diverse data into AI training is technically feasible but socially and politically complex. Organizations like Seva, working to legitimize Indigenous knowledge, face systemic barriers, from funding to validation. The challenge is not just to include more voices but to dismantle the hierarchies that devalue them.
As I reflect on my father’s herbal treatment, I realize that uncertainty is a starting point. We don’t need to blindly accept or reject traditional knowledge, but we must engage with it honestly and critically. The intelligence we need most is the ability to see beyond the hierarchies that define what knowledge matters. Without this, we risk erasing wisdom that could help us address global challenges, from climate change to sustainable living.
Here’s the final thought-provoking question for you: As we march toward a future dominated by AI, are we willing to listen to the voices we’ve silenced, or will we continue to erase knowledge systems that could save us? The choice is ours, and the time to act is now.
