In a stimulating conversation at the NXT Conclave 2025, organized by Bharat Mandapam in New Delhi on February 28, Smriti Irani, former Union Minister, discussed with Rishabh Gulati, Editor-in-Chief of NewsX. Discussing the role of artificial intelligence (AI) in gender equity, Irani insisted that technology is not devoid of bias and AI-based systems tend to sharpen existing social inequalities.
Technology and Gender Bias in AI
In her address, Smriti Irani highlighted the myth that technology is neutral. She stated, “I think one of the greatest fallacies that have been sold is that technology is agnostic to bias. As AI grows, one recognizes, realizes, and many people monetize the fact that technology has created certain schisms, certain gaps which have become opportunities for many those who have lived as thought leaders or as policy makers dedicated to equity. Their decadal presumption was that equality will be assured. That is still a challenge when you talk about the age, let’s say, of AI, many people compute how much it will add to a nation’s GDP.”
She emphasized that though the role of AI in boosting economic growth is talked about, nothing much is mentioned regarding how it impacts consumer behavior at a psychological level. Quoting worldwide studies, she mentioned that 80–90% of what women earn is spent on healthcare and education. Yet, AI-powered consumer algorithms push women towards spending money on perishables and not towards investment opportunities. “Research says 72% of purchasing decisions are affected by algorithms. One would presume that that algorithm would deepen your engagement as a woman on spans of education and health, because you are genetically predisposed to it. But what happens is that technology and the Journal of Psychology especially says that women, through algorithms, get 22% more spending on perishable consumables and at 30% less likely to get algorithmically driven data sets or content, which encourages them to invest.”
AI and Credit Bias: Berkeley Research Insights
Smriti Irani referred to a study by Genevieve Smith at Berkeley that had analyzed gender bias in 133 AI systems. The research found bias in 44% of the systems, which resulted in financial losses for women.
She gave an illustration of how Smith and her husband, who both had comparable financial histories, received different credit limits because of AI bias. “Genevieve Smith, along with her husband, applied for a credit card in the US. She had the same money in the bank. They in fact had a joint account. Her credit score was better than the husband’s, and her tax returns were as transparent as the husband’s. She got credit 10% less than the husband, not 20% less.” That led her to begin researching the technological impact of how women increase, or how women’s creditworthiness decreases over time.
Such is the real-life scenario that unveiled how AI-based financial systems tend to undermine women’s economic power, perpetuating imbalances that needed to be wiped out centuries ago.
Gender Disparity in Facial Recognition and AI Systems
Smriti Irani also identified racial and gender discrimination in AI-based systems. She underlined the fact that 80% of facial recognition technology is trained mostly on white skin, which generates flaws in identity verification for individuals with other ethnic backgrounds. Most AI models also base their safety tests on male physiology, which causes serious implications for women.
“They use a male body or the dynamics of a male body on an AI system. They do not look at anticipate a female driving and the impact on a female body and do not anticipate a pregnant woman driving. And research now has showed that in the same scenario, because women at the secure testing drive phase itself are missing as a consideration in research, 44% women are more likely to be injured in the same kind of accident that a man may have, and 17% more are likely to die as compared to men.”
She also showed how prejudices cut deeper than just face recognition. AI imaging devices find it challenging to identify left-handed handwriting due to the excessive predominance in training data regarding right-handed users.
“If it perpetuates bias, I mentioned left hand right, and you ask an AI imaging tool to come up with an image of a left hand person writing, they can’t, because it is such a bias that all the images they’re trained on are right handed. So when, there can be a racial bias when it comes to the data scraping that you’re doing, we’ve seen that in facial recognition systems, 80% of them are based on either a white skin or a male even, fingerprint scanners.”
The Economic Cost of Gender Inequality in AI
In response to whether gender bias in AI would have an economic cost, Irani identified that there is a silence surrounding the monetary loss arising from institutional biases. She demonstrated how attention has always been geared towards rights as opposed to the economic effect.
“I think one of the greatest tragedies which has happened on the issue of gender, is that we’ve always had a rights based conversation on it. If there was an equal amount of conversation on the economic impact of not ensuring equity, we would have not seen the kind of bias which right now is hidden in a lot of rhetoric.”
She emphasized that if the world had an understanding of the economic drawbacks of AI bias earlier, attempts toward achieving artificial general intelligence would have been more balanced.
“But we would have not seen it if we had made an evidence based economic case for it, maybe a decade, decade and a half ago. Okay, so huge amount of effort is now being made to reach the goal of artificial general intelligence, and we are going to be devoting hundreds of billions of dollars in the compute in the hardware to get this done.”
A Call for Inclusive AI Development
Smriti Irani’s address at the NXT Conclave 2025 underscored the urgent need to address AI’s inherent biases. She highlighted the disparities in algorithm-driven financial decisions, consumer behavior, and safety measures, urging for inclusive and fair AI development.
Her insights served as a wake-up call for policymakers, technologists, and industry leaders to ensure that AI works for everyone, regardless of gender or race.
