Tell me about your AI journey so far. What inspired you to carve out a career in data science?

I have been in love with maths, physics, and machines since childhood. Being an only child, my parents obviously had the biggest influence on my priorities, mindset and career choices. My mom had completed double MSc’s with university ranks, and had been a professor in sciences all her life. My dad, who was an accountant by training, had a brilliant mind that absorbed all great writers’ works be it in arts, literature or science. He never raised ‘just a girl’, but forced me to try and learn everything, from fixing lights and wires at home, to fixing radios and bicycles. Our favourite Sunday morning activities included solving complex math problems without pen and paper.

I wrote the highly competitive engineering entrance examination simply because I enjoyed writing maths and physics exams! I always wanted to study physics but when I got admission into one of the best engineering universities in the country, my dad put the idea into my head saying that engineering is basically applied maths and physics. He was right, again! I loved studying engineering, especially mechanics and machines, and came on top of the class full of boys. I enjoyed working in the shopfloor, in my first job at Tata Motors Jamshedpur, and also got interested in software development.

Then again, thanks to Dad, I took all AI and machine learning courses during my MS at Strathclyde, way back in 1997-98. I got a chance to put all my learnings into action when I started working at GEC Glasgow to build an expert system- first of its kind in the entire industry those days. I learnt so much in those 2 years at GEC that it became my natural choice in the technology domain, for a PhD.

What's your area of expertise in AI and why did you choose this?

I had this fascination towards languages and literature, while also loving maths and sciences. Now, at the wrong side of 40, I get to realize that it’s alright to love both literature and maths! Human languages are still the biggest unsolved problems in AI. So that remains my area of interest, stemming from my GEC project in 1997-98 when I codified the experiential implicit technical knowledge of 1000’s of engineers into an expert system that could recommend resolutions to operational and design problems. We didn’t have all these python libraries back then, and had to code everything from scratch, using logical reasoning techniques, forward and backward chaining, processing regular expressions in C, C++, LISP, PERL, even JavaScript! But, that way, the learning had been solid, and stayed with us forever.

During my PhD work, I single-handedly built all the system prototypes for knowledge modelling and knowledge entropy measurements. It made the entire process so rigorous and exhaustive, but I loved every moment of it. I am still working on the language processing side of AI, and few of my patents focus on building explainable language-processing AI solutions. Checking biases and ensuring fairness in language corpus are critical governance aspects of AI in enterprises and society, as the solutions go into production. My experience and expertise have therefore become increasingly important since past few years.

What challenges have you faced in creating a niche for yourself in the field of tech?

The tech field of AI and data sciences is extremely conventional and male-dominated. Even in global companies, implicit biases create terrible mess. Female tech talents in AI are often not taken seriously enough, even at leadership levels. I have been lucky in some cases, with male bosses having open, learners’ minds, e.g. at Gartner and Wipro. But there have also been instances of male peers condescendingly commenting on how my career ‘shouldn’t matter much’, given I don’t have any financial liability or insecurity, because my husband is in a good corporate position.  

We ourselves as women are also not free from prejudices and biases. First, we have to forget that we are from a different gender, it’s not like we’re some different species altogether. Whenever anyone gives unsolicited advice, being direct in pointing it out and not accepting any type of non-peer like condescending behaviour, are must-do’s. We all must have the humility to learn from each other. If I respect and learn from you, I expect the same attitude in return, irrespective of gender, race, colour or geography.

How will women in tech roles help mitigate bias and promote inclusion?

We need to accept the reality and constraints, that the reality won’t change by itself, or change in a day. We need to practice what we preach to our client teams- on strategy and change management. People may not take us seriously at first. But, basis my personal experience of working as a consultant for 20+ years, within 2-3 minutes into a client conversation, the smart ones get it and then they forget your gender and just love you for your content and ideas.

Throughout my work-life, my client leaders and teams have become my best friends, and this trust network has nothing to do with the brands we work for. It’s more about the personal brands we create, with confidence, with our passion and ability to propose completely audacious, out-of-the-box ideas and our courage to challenge the status quo and actually make those ideated changes happen! Finally, when we can show the client teams what they have achieved with us as partners, in hard numbers and dollar terms, they trust us with their future, career and businesses, irrespective of our gender or colour or accent. Numbers always work- they have no colour or gender.

Another strategy that worked very well for me is to build a very strong tech band of boys and girls (if and when they are available) as my juniors. I have never cared much about hierarchies. To break glass ceilings, we have to be the change we want to see and break the ‘seniority’ glass ceilings ourselves. Learning new algorithms and packages, staying hands-on, sharing the enthusiasm, studying new papers together with bright kids 20 years younger than you- all these informal interactions keep you young in mind and keep your practical AI knowledge relevant and fresh. This worked best for me.

What do you think are the biggest limiting factors for women not to advance their careers in tech, esp product development? What can change?

The first and foremost hurdle, in my personal experience, is our own insecurities and lack of confidence- not giving our ideas a chance. From ideas to appearance, most women are always looking for external validation. At the risk of stereotyping and sounding rather judgemental towards my own gender, I often feel we are far less outspoken than our male peers, and are risk-averse by default. Many of us have deep enough technical acumen but aren’t daring enough to speak out our minds and challenge the status quo. We tend not to speak up even when we know that we know better- this is especially true for science and tech subjects.

Secondly, prioritizing family over work is a big issue. Why can’t we spread the passion for what we love (e.g. AI, data science, maths, stats, eco, linguistics) amongst our kids? I have always discussed work with my two girls, and thank God that I always respected their intelligence and still listen to their suggestions at all times.

Finally, bothering too much about how others perceive us is a huge mental block that many of us suffer from, knowingly or unknowingly. We ‘appear’ as ‘overly approachable’ if we show enthusiasm and passion about our domains, or ‘aggressive’ if we demand what we deserve. No matter what we do, may times we are judged by folks, male or female, who haven’t even achieved half of what we have and still feel entitled to give us precious advice.

Based on my experience so far, not listening to people stupider than us is the best principle to utilize our own potential and merit. We waste far too much time to get approval from others. We must let our work speak for us, and let our passion drive us. If some people keep judging us or don’t approve, that’s their insecurity and their problem, not ours. We will gently point it out to them as it’s our duty to be vocal about unfairness. We must not accept blind, tact-less and fact-less criticisms. But we will not waste our precious time and energy arguing with men on what women should do. We must show ‘confidence’ confidently, that yes we know what we should do and don’t need any unsolicited help or advice. Our brains are just as good!

How do you think corporates are moving the needle in terms of supporting more women to participate in tech building/development?

There are initiatives taken up by several government agencies and institutions, to promote digital literacy among girls, enhancing digital access and adopting digital technologies with special attention to the digital gender divide. For example: The 25 by 25 goal that G20 members committed to in Brisbane, EQUALS and G20 #eSkills4girls initiatives. But much more needs to be done. At the same time, women also mustn’t try to take advantage of gender-specific promotion programs and policies and try to take the easy way. Equality doesn’t mean more privileges. Discriminatory privileges create more grudges, misconceptions, mistrust and more disparity.

We need to strive to be at the top of the ladder, in core tech stuff. And we need to build confidence that yes, we can. If we don’t find enough role models in AI or core STEM domains, we can borrow them from other adjacent domains. Formula for success is the same for women- talent/ merit, and passionate hard work. For example- if you love AI, then working on it for 15 hours a day isn’t a stressful at all. It can be your hobby! It’s just as relaxing as doing maths in between preparations for history exams next day- a strategy my girls have invented and taught me.

What's the one thing that you see AI transforming completely?

I have this dream about singularity that- someday- maybe in our children’s lifetime- super AI will help us humans become a better species. If we the AI practitioners and researchers work tirelessly on eliminating data biases and algorithmic biases, towards building fair and ethical, socially responsible AI, then my definition of super AI will definitely become a reality. Participation of more women in AI tech research is all the more relevant now, as we must have women questioning and building data and algorithm checkpoints.

We need to consider a hidden upside of AI biases- thanks to AI usecases being questioned in terms of fairness, the intrinsic human biases are getting detected! For example- the COMPAS project actually revealed data-driven patterns of biased judgements.

AI will not only increase longevity or lifespan, by giving old-age bot assistance or pharma drug discovery n simulation, but it will improve quality of life in a ubiquitous way if we train it right. Algorithms and math models are intrinsically rational, so AI will show back to us our own irrationalities in decisions and actions and help us question and challenge our own thought processes. This way, AI applications have the potential to make us humans more rational and considerate, who are open to factor in others’ viewpoints and optimize societal value as a whole e.g. ESG, green AI, impact investments and social AI, with AI for governance and governance for AI.

Your biggest AI nightmare?

My scariest AI nightmares include usage of AI in warfare, and bad training data making bad AI solutions and people trusting bad decisions due to over-expectation and false confidence on AI models’ outputs. The Big AI divide is becoming increasingly daunting, between the data-have and have-nots - people with access to good data and algo’s and skills and people without. AI-native generations may actually increase this divide.

The other nightmares of course involve data security and decision security and trust issues, e.g. sensitive data vs. security trade-off.

What's your advice to women who want to embark on a journey similar to yours?

Per some surveys, women have been shown to have relatively better skills at multi-tasking and balancing multiple viewpoints/ perspectives. This is an essential requirement in AI solutions design. So, it’s imperative for more women to participate in the technical architecting and designing of AI solutions (not just ML or python coding).

We must strive to get into core tech- e.g. algorithms building, metrics and measurements of fairness, risks and ESG. We must execute on our ideas and own patents. We must move from peripheral functions like marketing, communications, talent management for AI, training & upskilling, program management etc.

Patents are actually highly effective in bringing women tech talents to the limelight, given that patents are measurable objectively by numbers and impact and everyone trusts numbers!

We the women AI practitioners must get into problem solving and multi-pronged solutions design, while building new skills around multi-dimensional roles e.g. for culture change. Women are not just diversity cosmetics; women can actually handle diversity better. This century is driven by AI and women. When the two key change levers are added, drastic, robust and successful changes in society for good are inevitable. Government initiatives on AI also must factor these aspects in, and prioritize women-led initiatives on AI especially on governance best practices and holistic impact investments that build social equity.