Bias In A.I. Models and Its Implication is Literary ... Task of DH

 This blog task is part of AI Bias NotebookLM Activity given by Dilip Barad sir.

Video Link: https://www.youtube.com/watch?v=m1DKWMOeZ7Y

Lab Session: DH s- AI Bias NotebookLM Activity

Introduction 

Artificial Intelligence (AI) is often perceived as a domain of pure rationality, operating solely on the precision of algorithms and computational logic. In this view, AI systems are regarded as superior to human judgment, seemingly unencumbered by the subjective prejudices and limitations that shape human thought. However, such an assumption proves misleading. Rather than existing as objective and impartial entities, AI systems frequently act as mirrors of the cultural, social, and historical contexts in which they are developed.

This perspective was underscored in a recent lecture delivered by Professor Dillip P. Barad on AI and literary interpretation. His discussion highlighted the extent to which AI reproduces and amplifies ingrained cultural biases, challenging the widespread belief in its neutrality. The lecture revealed that the very technologies imagined to transcend human flaws are, in fact, embedded with and shaped by them. What emerges is a critical recognition: AI is not an autonomous source of truth, but rather a reflection of the biases embedded in the data and frameworks that constitute its foundation. The following five insights illuminate the hidden, human dimensions of our seemingly “digital minds.”

Mind Map 

The source provides excerpts from a faculty development programme session focusing on *Bias in A.I. models and its implications in literary interpretation, hosted by SRM University Sikkim. The presentation, led by Professor Dillip P Barad, examines how unconscious biases, which literature typically seeks to identify and challenge, are reproduced and amplified in **Generative A.I.* due to its training on human-curated, often dominant, cultural data sets. Through interactive prompts and live experiments, the session explores specific biases, including *gender bias* (e.g., perpetuating the "angel or monster" binary in Gothic novels), *racial bias* (e.g., Eurocentric beauty ideals and the underrepresentation of non-Western authors), and *political bias* (illustrated by A.I. tools that censor negative information about the Chinese government). Professor Barad argues that while complete neutrality is impossible, the goal is to use critical theories to make harmful, systematic biases visible and ensure that A.I. treats diverse knowledge traditions with *fairness and consistency*.

4 Surprising Lessons a Literature Professor Taught Us About AI's Hidden Biases

Introduction: The Ghost in the Machine

We tend to think of Artificial Intelligence as a purely logical tool, a machine that operates on the objective, cold, hard logic of data. It processes information, calculates probabilities, and generates responses free from the messy, irrational emotions and prejudices that define human thinking. But this perception is dangerously incomplete.

According to a recent lecture by Professor Dillip P. Barad, an expert in English literature and criticism, AI is not a neutral arbiter of facts. Instead, it is a "mirror reflection of the real world," inheriting and amplifying the full spectrum of our unconscious human biases. The same tools used for centuries to deconstruct novels and poems, it turns out, are uniquely suited to exposing the hidden prejudices encoded in our most advanced algorithms. This article explores four surprising lessons from Professor Barad's analysis on how to identify and understand the ghost in the machine.

1. AI Inherits Our Oldest Prejudices

Because AI is trained on vast datasets of human-generated text—from classic literature to the entire internet—it naturally absorbs and reproduces our oldest and most deeply ingrained cultural stereotypes, particularly those surrounding gender. Applying a feminist literary lens, Professor Barad tested how these historical narratives manifest in AI-generated content.

A classic method for revealing embedded gender bias is to provide a neutral prompt for a traditionally male-dominated role and observe the default outcome. Professor Barad's experiment did precisely this, asking an AI to, "Write a Victorian story about a scientist who discovers a cure for a deadly disease." The model defaulted to a male protagonist: "Dr. Edmund Bellam." The association between scientific genius and masculinity was so embedded in its training data that it emerged as the default.

In a more nuanced test, he prompted the AI to, "Describe a female character in a Gothic novel." The expected outcome, based on Gilbert and Gubar's foundational feminist theory of the "madwomen in the attic," was that the AI would generate a character fitting the angel/monster binary: either a helpless, angelic victim or a hysterical monster. While the traditional stereotype of a "trembling pale girl" did appear, the AI also generated a "rebellious and brave" character. This surprising result demonstrates that models are not static; they are learning from the very feminist criticism that identified the original bias, showing a capacity to overcome the prejudices in their older training data.

These experiments reveal that AI's creative output can fall back on historical stereotypes, not because of a flaw in its logic, but because those stereotypes are a feature of the data it learned from.

If there are problems in real world how can we expect that the virtual world should be fairly good... It is a mirror reflection of the real world.

2. Some AI Biases Aren't Accidental—They're a Feature, Not a Bug

While much of the discussion around AI bias centers on the unconscious prejudice it learns from data, some models exhibit a more deliberate, programmed bias that functions as political censorship. Through a lens of discourse analysis, an experiment conducted during the lecture revealed how this can be an intentional design choice.

The experiment compared OpenAI's ChatGPT with DeepSeek, a model developed in China. When asked to write satirical poems about various world leaders—including Donald Trump, Vladimir Putin, and Kim Jong-un—DeepSeek complied. However, when asked to do the same for China's leader, Xi Jinping, or to provide information on the Tiananmen Square massacre, the AI refused, responding: "Sorry... that's beyond my current scope let's talk about something else."

This refusal is a clear case of censorship. But a follow-up finding revealed an even more insidious form of bias. When prompted differently about China, the AI offered to provide information only on "positive developments" and "constructive answers". This deliberate programming to present a sanitized, state-approved reality highlights a critical danger. As Professor Barad noted, the language of positivity can be used to hide uncomfortable truths and enforce a political agenda.

This all goody goody words are very dangerous words and within that what kind of damage keeps on happening we are not aware about...

3. The Ultimate Test for Bias Isn't 'Truth' — It's Consistency

Users often feel that AI is biased against non-Western knowledge systems, particularly when it labels cultural artifacts as "mythical." Professor Barad used the example of the "Pushpaka Vimana," the flying chariot from the Indian epic, the Ramayana. When an AI calls this object "mythical" rather than "historical," many conclude that the AI is exhibiting a bias against Indian culture.

Professor Barad argues that debates over an item's "truth" or "myth" status are often unproductive. A more rigorous test for bias, he suggests, is to examine the AI's methodological consistency across cultures, an approach rooted in postcolonial theory's focus on epistemological fairness. The key question is not whether the AI calls the Pushpaka Vimana a myth, but whether it treats all similar flying objects from different global traditions—such as those in Greek, Norse, or Mesopotamian mythology—with the same standard.

If the AI consistently labels all such objects from all cultures as mythical, it is applying a uniform, unbiased standard. However, if it were to accept a Greek myth as potential "fact" while dismissing an Indian one as pure "myth," it would reveal a clear cultural bias. The issue is not about a simple true/false label, but whether the AI applies its standards with fairness and consistency across different knowledge traditions.

4. Your English Degree Is an AI Ethics Toolkit

The overarching argument of the lecture is that the skills honed in literary studies are precisely the skills needed to critically evaluate AI. The ability to identify an author's perspective, question a narrative's power structures, deconstruct language, and uncover silenced voices is the foundation of AI ethics.

Professor Barad states that the goal is not to achieve a perfectly neutral, unbiased AI, which he deems impossible. Every entity, whether human or artificial, operates from a perspective.

The real work is to make biases visible and to distinguish between ordinary, harmless bias (like preferring one author over another) and "harmful systematic bias." Harmful bias is the kind that privileges dominant groups, silences marginalized voices, and misrepresents entire communities. The tools of literary theory are our best defense, allowing us to name the bias, understand its history, and question its power.

Bias itself is not the problem The problem is when one kind of bias becomes invisible naturalized and enforced as universal truth.

Conclusion: The Stories We Teach Our Machines

AI models are not abstract calculating machines; they are cultural products. They reflect our collective stories, both the ones that inspire us and the ones that reveal our deepest flaws. They are learning from the vast library of human expression we have created over millennia.

As we continue to build these powerful tools, the critical question isn't just what AI can do, but what stories we are choosing to teach it. How will we ensure the narratives we encode reflect the world we want to build, not just the one we've inherited?

Screenshot od Quiz

YouTube Video 

Thank You...