Musings-2

 

I came across a reel recently. I don’t know what to make of it. I consider myself too backdated, backward minded, conservative, possibly jealous(sic). The reel concerned Suzanne khan, Hrithik Roshan’s ex-wife’ s Christmas party this year. The handsome hero Hrithik was there with his current girlfriend, his ex-wife, here Suzanne with her current boyfriend, the two lovely kids of the couple and other guests. How naturally they all clicked pictures together, Hrithik, his current, Suzanne, her current, their parents, the boys – all in one frame.

I imagined my father’s face had it been such a situation in my house. He would have shouted, openly slanged the nuisance I made of his upbringing and vehemently ensured that he had nothing to do with it. My mother-in-law would have kept mum, but ensured that she was absent from the whole arrangement. It is impossible to imagine the hearty participation of anybody else – I can see grumpy, sad, heart broken faces lay aside smiling and clicking pictures or that compulsory photo sessions in a Christmas party.

 

I now come to believe that I am extremely poor, I have a very low, backdated mindset. I am not comfortable sharing my husband or seeing him cozy-cooing with some other women in my presence. It requires a lot of mental strength. Even after a divorce, or if moving on has been smooth, I wonder how many of us would be comfortable in such gatherings. Was it only an hour visit? Did they not feel uncomfortable at the end of the party? How easy is it to hug an ex-husband’s current partner and putting it up with a smiling face? I believe it requires a lot of grit, patience, strength, none I think I would have in such a situation. I don’t want to be judgmental, I haven’t seen the world as they have, I haven’t seen stardom and never will, but I do admire how easily they can mould themselves to fit the box perfectly.

 I don’t think I will (who knows, “kal kisne dekha hain”). I blame it on my middle-class upbringing, middle class values, high notions of loyalty, fidelity, commitment and compromise (notions that in the modern days are constantly under the questionable radar). I saw my parents together- loving, caring, quarrelling – staying put for 65 years (until my mother’s death) and believed this to be the norm.

Things are changing today, and we advise our juniors and children that it is better to get separated rather than live in an abusive relationship. But this particular reel vexed me. How far have we come!

I would rather be poor for now! :)

Musings -1

 

My mind is in a constant state of flux. It never is still. I can always feel its continuous movement, I can never be free. Sometime back I thought may be these are the first signs I might go crazy, but now I have learned to live with them. Then again, I thought these thought processes of mine could be, yes 'COULD BE’ channelized into something inspirational, something good but the inadvertent interruptions in my life in the form of my workplace duties, or my household demands makes my thoughts go awry and before I could drop in that ‘high order thinking’ into a box I am lost.

I don’t write; I like to read. I heard most people like to read; I am not sure. So, I say I heard. When I was a young girl in my 20s, my mother and I spent endless evenings just talking. Just as I like to talk, I like to listen to people. Life experiences interest me. I don’t like gossiping about people or check on what somebody had for breakfast, lunch or dinner but if that one particular dish had a special spice which your grandmother liked and would make, I would hear about it. Much later in life I realized I was an observer. I can spend hours staring at things, thinking. I remember how I would look at the distant mountains from our 4^th floor apartment in Imphal and emotions would overload me. I would start crying for reasons unknown. As a student of literature, I learned these were my ‘spontaneous overflow of powerful feelings’. Had it not been for Wordsworth I would have never had an expression for such a feeling.

I regret not learning my mother tongue properly. We grew up in a different State, and learned a different language as our vernacular. Later, I took upon myself the arduous task of learning the language in which I dreamt. I think it is important to know how to read, write and speak one’s mother tongue. It is important that one should at least read the literature written in one’s language. I regret being a slow reader. My husband loves to read too. I would ask him to read. When my daughter was a child, he would read the vernacular classics and we would all take the pleasure of that story telling. I am reminded of the Grandmother’s Tales – our ‘thakurmar jhulis’ – a granny with a bag full of stories. In our vacations we would read Panchatantra, Jataka Tales, Anandomela, Chandamama , Tinkle comics, Mandrake, Phantom, as much as we read Wisdom, Tales from the Bible, Amar Chitra Katha, Aesop’s Fables, Tales from the Arabian Nights etc. We enjoyed our vacations playing, reading and resting. Now children plan online games and plan outings with their friends.

As individuals, not all of us can say what we want to say but sometimes it is important to let people know what you think, for those who actually matter. I loved my mother and I still do. I was harsh to her when I could not convince her that the world was not as simple as she thinks it is, that your closest ones can be chameleons in disguise and most often than not people shed crocodile tears. But when she suddenly felt sick , something in my heart told me my mother’s days were numbered, and finally I told my mother…Maa, you are the best mother in the world…you did everything for us, no one could have raised us better.” I still don’t know why those words came to me, I saw my mother happy, smiling.I did not speak to her again. My mother left us.

Today I live with her teachings. She lives through me. I would do nothing that would disrespect her upbringing. My respect for her has doubled boundlessly. She would tell me ‘Don’t stress yourself about things beyond your control. When you cant do anything, just don’t think about it. Just let things flow.” This day, when my voice chokes in my throat, as I remember my mothers sweet face , I stop thinking. That calms me.

Of late, I realized I am not advancing towards dementia, alzheimers or some other age related mental problem(I rather use ‘ mind-related, mental is a big word), I just have a thinking brain and to keep it engaged I need to do constructive things. For instance ‘Read’.

Is AI gender biased? Understanding ChatGPT

 

It all started when I was preparing a PowerPoint presentation on ‘Gender Sensitivity’ I asked ChatGPT to give me a picture where one partner was cooking and the other was making a chair. The first picture that ChatGPT gave me was of a man finishing the chair and the woman cooking. I asked it to change the roles, make the man cook and the woman do the chair. He helped me with the picture but something else changed too. The lady became tomboyish, had short hair, wore trousers. I asked the app to make the lady wear Indian clothes and hairdo but my problem was not resolved. Her hair still remained short and she looked masculine. I was confused and bitter…why did such a simple set of instructions appear so difficult to such a technologically driven application?

I planned a few questions and asked directly :

IS AI GENDER-BIASED?

And the answer☹(ctrl+c/ctrl+v)directly from chatGPT

It can be — but not because AI is inherently sexist or has a personal agenda.
The bias comes from three main sources:

1.      Training Data Bias – AI learns patterns from huge datasets, many of which are scraped from the internet, books, media, and other human-made sources. If those sources have gender stereotypes (e.g., associating men more with leadership and women with caregiving), the AI can unintentionally replicate them.

2.      Algorithmic Design – Even if data is balanced, the way algorithms weigh, classify, or predict can magnify small biases. For example, a hiring AI might favor male candidates because the historical company data showed more men in senior roles.

3.      Human Feedback Loops – AI models often improve through user feedback. If that feedback carries bias (consciously or unconsciously), the system can reinforce it — a bit like a rumor growing stronger the more people repeat it. In short: AI reflects our world back at us, warts and all. That’s why researchers now spend a lot of effort on debiasing techniques, ethical guidelines, and more representative datasets — but no system is 100% bias-free yet.

 

ME: Can you show me specific references where AI is gender-biased

Here are specific, documented examples and references showing how AI systems can exhibit gender bias — backed by research, real-world cases, and academic studies:

---

1. Bias in Hiring Algorithms

Amazon’s AI Recruiting Tool

• Amazon’s automated hiring system learned from past resumes, which were mostly from male applicants and engineers. The AI began penalizing resumes from women, especially those including all-female educational histories or women-specific terms, leading to gender bias in candidate scoring. It was eventually abandoned because of this bias. Digital Adoption

LinkedIn Job Recommendation Systems

• Research has shown LinkedIn’s AI algorithm tended to favor male candidates over equally qualified women in job recommendations, reflecting systemic bias in job search ranking tools. Crescendo.ai

Resume Ranking Bias by LLMs

• A study on state-of-the-art language models found AI tools favored male-associated names far more often than female-associated names when ranking resumes, demonstrating bias in gendered hiring relevance. washington.edu

---

2. Gender Bias in Image Generation and Representation

Stereotyped AI Images

• Image generation models like Midjourney and DALL·E often portray occupations in gender-traditional ways — e.g., accountants and professionals as men, and caregiving professions as women — reinforcing stereotypes. TUM

Generative AI Bias Against Women

·         Academic research on generative AI tools (Midjourney, Stable Diffusion, DALL·E 2) found gender bias in isual depictions: women were portrayed with stereotypical expressions (e.g., smiling more) and biased age/facial features compared to men, suggesting gendered portrayals. arXiv

---

3. Language and Translation Bias

Machine Translation Defaults to Male

• Research on gender bias in machine translation (e.g., Google Translate) shows the system often defaults to male pronouns or male occupations (like “engineer”) when a translation could be gender-neutral, reflecting and amplifying societal bias in training data. arXiv

---

4. Facial Recognition Performance Differences

·         Facial Recognition Systems (Gender Shades) The Gender Shades study found commercial facial recognition tools (e.g., Amazon Rekognition, IBM, Microsoft) performed better on male faces than female faces — with especially high error rates for dark-skinned women, highlighting intersectional gender and race bias in AI. Wikipedia

---

5. Structural and Systemic Bias in AI

Bias Reflected Across AI Systems

• Gender bias isn’t limited to one application; AI models trained on societal data tend to mirror existing gender inequalities (e.g., associating “nurse” with women and “doctor” with men), which can widen gender gaps if unchecked. Consilium

UN Women on AI Reinforcing Gender Inequality

• According to UN Women, AI systems — from hiring to healthcare to governance — can amplify pre-existing gender biases, further disadvantaging women if not proactively corrected. UN Women

Underrepresentation in AI Development

Participation Gap

• Studies highlight that women are significantly underrepresented in AI development and at AI conferences, which contributes to biased systems because fewer female perspectives shape the design and testing of models. UNESCO

---

Why This Happens

• Bias in Training Data: AI learns from text, images, and decisions that reflect historical gender inequality, so it reproduces those patterns. The Gradient

• Lack of Diverse Development Teams: Fewer women in AI roles means less scrutiny of gender issues during development. Premier Science
• Algorithm Design Limits: Even with balanced data, how AI measures relevance or similarity can still implicitly favor male-associated patterns. The Gradient

---

 

• Systematic reviews on AI gender bias, summarizing detection and mitigation. M-Chair
• Machine Translation Gender Bias (Prates et al.), experimental evidence of gender defaults in translation outputs. arXiv

---

So, if you are still wondering and saying aloud ‘REALLY??’

Yes! It is ‘Really’ gender-biased. We may have come far but for gender equality in the truest sense we still have a far way to go.

For real progress to happen, change should take place at the base, that is at the basics. Because of the years of patriarchal domination in the different fields, subordination is so embedded in the system that it is difficult to remove its traces even from a system generated app.

However, things are changing, we are now comfortable seeing our mothers read and fathers wash dishes. Looking at it the other way, men need not always be the bread-earner, women can be. Our eyes are still not set to it, but it has to be. Empowerment of women will come only when we learn to respect her opinion, choice and thinking.  Negligence, domination and subordination have pushed women so far in the background that to be equal she needs that small advantage – in the form of supportive hands – liberal, progressive man, government and the society.