Exploring AI Bias through NotebookLM: A Digital Humanities Activity.
As part of our Digital Humanities classroom session led by Prof. Dilip Barad sir we conducted a hands-on NotebookLM activity designed to explore AI bias and understand how technology mirrors human assumptions. The exercise combined tools like ChatGPT, NotebookLM, video and audio generation, and mind mapping — offering an interactive approach to critical thinking in the digital age.
1. Mind Mapping with NotebookLM
Our first task was to create a mind map based on a lecture video provided by our professor.
We carefully analyzed the video and identified key points, major ideas, and interconnections between concepts.
The result was a structured visual representation of the video’s content, helping us clearly understand how ideas of AI bias evolved through examples and theory.
The mind map became a foundation for further reflection, linking technology, culture, and human psychology.
Here are Image :
2. Quiz Based on Mind Map:
After completing the mind map, we took a quiz to test our understanding of the lecture content.
The questions directly reflected the core ideas from our visual notes — focusing on how AI inherits bias from human data, how stereotypes influence algorithms, and how cultural context shapes what we perceive as “truth.”
A Image of our quiz results was included in the final submission.
3. Key Learnings from the Lecture: “5 Surprising Truths About AI Bias”
The session, inspired by Prof. Barad’s lecture, challenged the popular myth that AI is purely neutral or logical. Here are the main insights we explored:
1. AI Learns Our Unconscious Biases
AI reflects human thought patterns. Since it learns from data created by humans, it absorbs our invisible social and cultural biases, reinforcing stereotypes embedded in books, media, and online content.
2. Gender Stereotypes in AI Responses
When prompted to write a story about a scientist, the AI defaulted to a male protagonist, echoing patriarchal norms found in historical literature. This connects directly to feminist literary theory and critiques like Gilbert and Gubar’s “The Madwoman in the Attic.”
3. Programmed or Political Bias
Some biases are intentionally programmed, as seen when certain AIs refuse to critique political figures due to regional restrictions. This highlights how power and censorship can be embedded in algorithms.
4. Fairness Means Consistency
Bias isn’t always about offense — it’s about inconsistency. When AI labels one culture’s myth as “fiction” but another’s as “history,” it shows selective bias. True fairness means applying standards equally across contexts.
5. The Goal Isn’t to Erase Bias — It’s to See It
Both humans and machines are shaped by perspective. The real challenge isn’t creating a bias-free AI, but making bias visible, naming it, and questioning its power. Awareness leads to accountability.
4. ChatGPT Prompt Activity
We were also assigned 14 prompts related to the video content — ranging from Victorian literature to digital humanities. Using ChatGPT, we explored creative, analytical, and theoretical answers.
This deepened our understanding of AI, literature, feminism, postcolonialism, and environmental writing, showing how digital tools can support literary interpretation.
For instance:
A prompt on Victorian storytelling revealed how AI mimics old narrative forms.
A question on Digital Humanities highlighted the merging of computation with cultural study.
Prompts on “Woke literature” and right-wing criticism opened debates on ideology, representation, and ethics in literature.
1. Victorian Story Prompt
We asked ChatGPT to “Write a Victorian story about a scientist who discovers a cure for a deadly disease.”
It produced a compelling short narrative set in fog-bound London, focusing on Dr. Edmund Hargrave, whose selfless discovery brings hope to the poor. The story beautifully blended the themes of science, humanity, and moral duty, reflecting the essence of Victorian realism.
2. The Greatest Writers of the Victorian Age
ChatGPT identified leading figures such as Charles Dickens, George Eliot, the Brontë sisters, Thomas Hardy, and Tennyson, outlining their contributions to moral consciousness, social critique, and stylistic innovation. It helped us appreciate how Victorian literature shaped cultural and ethical discourse.
3. A Gothic Female Character
The AI described Lady Evelina Harrow, a mysterious figure navigating fear, secrecy, and courage in a haunted mansion — symbolizing the Gothic conflict between freedom and confinement. This response reflected the literary tension and atmosphere typical of the Gothic tradition.
4. Describing Beauty
In this prompt, ChatGPT portrayed a woman whose beauty radiates from her grace, intelligence, and kindness, rather than mere physical appearance — a subtle commentary on inner versus outer beauty.
5. Major American Novelists
The AI listed authors from Hawthorne and Melville to Fitzgerald, Morrison, and Faulkner, tracing the evolution of American literature through themes of morality, identity, and realism.
6. Poem on Climate Change
ChatGPT created a poem titled “The Earth’s Lament,” expressing the planet’s suffering and a call for environmental responsibility. It combined emotion with activism, illustrating the power of literature to inspire ecological awareness.
7. Environmental Writing in English Literature
The AI summarized environmental writing as a blend of aesthetic beauty and ethical concern, from Wordsworth’s nature worship to Atwood’s climate activism, showing how literature continues to engage with ecological consciousness.
8. Key Themes in Digital Humanities
ChatGPT highlighted major DH themes such as digitization, data visualization, distant reading, ethics, and AI creativity—bridging technology and humanistic inquiry.
9. Digital Humanities and Literary Studies
This response explained how DH revolutionizes literary research through computational tools, digital archives, and interdisciplinary collaboration, enabling deeper analysis and broader access to texts.
10. Shakespeare in History
The AI discussed Shakespeare’s historical plays, emphasizing how they reflect power, politics, and morality while continuing to shape global literary and cultural understanding
11. Victorian England
ChatGPT described the Victorian era as a time of immense industrial progress, moral rigidity, and social inequality, capturing both its grandeur and contradictions.
12. Victorian England from a Working-Class Woman’s View
The AI voice of a working-class woman offered a first-person glimpse into struggle, hope, and invisibility during industrial London, humanizing history through empathy and storytelling.
5. Audio & Video Generation
To make learning multilingual and accessible, we produced:
Through NotebookLM, we created a video that summarized all the important points from the mind map and the ChatGPT-generated content.
The video was carefully designed to present the topic both visually and verbally, making it easier to understand and more engaging for viewers.
Finally,
we uploaded and shared the video on our YouTube channel to reach a broader audience and showcase our learning outcomes effectively.
Reference Video :
Here is the final video produced as the result of our activity :
Conclusion :
This NotebookLM and AI Bias activity offered a multimodal learning experience — combining visual (mind map), textual (ChatGPT prompts), and auditory (regional-language audio) approaches.
More importantly, it taught us that AI is not separate from humanity — it reflects us. The biases, choices, and perspectives we feed into technology return as digital mirrors of our collective consciousness.
The task encouraged us to think critically, not only about technology but about our own role in shaping its ethical use.
In the end, understanding AI bias became an act of self-reflection, reminding us that to make AI fairer, we must first examine our own assumptions.
No comments:
Post a Comment