Beena Ammanath is one of the most prominent names in the global AI ecosystem. Beena, currently the Executive Director at Deloitte's Global AI institute, is one of the outstanding AI & Data Science leaders who had worked across numerous industries such as IT, finance and e-commerce, for reputed companies such as Deloitte, HP, GE, Thomson Reuters and the Bank of America. She is also the founder and CEO of Humans for AI- a nonprofit organization focused on increasing diversity in AI.

Beena has been one of the most important voices on AI ethics and trustworthiness in AI models, and she is launching her book Trustworthy AI: A Business Guide for Navigating Trust and Ethics in AI on 22nd March, featuring an AMA with the author herself. INDIAai caught up with her to understand more about her work and vision in AI and ethics.

Beena, you grew up and studied in India in an era where digital technologies were nascent. You are one of the exceedingly rare people who studied data science in the early 90s and started a successful career in that field very early. What motivated you to choose this particular field back then? 

Early in my student days, I realized that I was very good at mathematics. Hence, it was logical for me to choose computer science, a new area for learning in India back then. I was part of the third batch or so, to have graduated with a focused degree in computer science. And not much was known about the field back then. My parents were certainly not very happy as it was not the typical 'engineering or medicine.' 

I really wanted to learn computer science as it seems to have a lot of subjects around maths, thinking that I would cruise through college without having to study much and fortunately, it did turn out that way. 

During those days, I realized that I was good with databases. We had a subject around advanced database systems. We are talking about DBS, Lotus and FoxPro, which I don't think many of your readers will remember. And when I got my first job, I latched on to the fact that I wanted to focus on the data space because I seemed to have an instinct and infusion to work with it.  

I started out as a SQL developer and kept my focus on the data aspect. I certainly didn't plan it, and now looking back, I did the right thing because I focused on my strength. And my strength was in understanding database and database systems. 

Did you think back then that data science and AI would become what it is today? 

Not at all! I studied back in college, mainly in textbooks, artificial intelligence, and LISP. And I remember during those days in 1995 or so, we would hear about all advances happening, and we used to discuss that in future when you are driving on the road, you're going to see electronic ad banners, which will be personalized for you.

Back then, there were no electronic banners, and it seemed so much in the future that we also felt that's never going to happen. So we were just imagining things. But now we live in a world where even talking to you about this comes across as we were limited in our imagination at that time. 

Every time technology advances, we try out a few things, and then we figure out there are more things you could do with the exact same technology. So it was sheer luck that I chose data, and it is one of those areas that has exploded. And now, I am glad to be able to build AI solutions and lead AI teams at companies where I am bringing value with AI and data. I could have never imagined this.

Coming to India, we are a country where the traditional industries are predominant in our economy, with almost 50% of our workforce engaged in agriculture, contributing 18 to 20% to the GDP. Considering that we have a vast youth population, how do you think AI adoption and automation will affect the country, its people, and the economy? How we can ensure that we will come out as a winner in this 'AI arms race'? 

Traditional industries will lag behind because we have never approached it from a data perspective. I'm betting on the youth of India to figure it out as not only the youth but everyone has a smartphone they're all familiar with. The digital literacy level has definitely gone up compared to 10 years ago. And if there is a little bit of incentive for digitizing the farms to some extent, adding sensors and doing some foundational work where you can start capturing data, it will get us further ahead.

We should never approach this as an AI problem that needs solving. Instead, we should look at it as how do we make it easy for people working in traditional industries to capture data and engage that audience so that we can set up the data pipeline. Once we solve the data pipeline, the AI part is actually easy. 

So, the more you can incentivize the youth, companies or entrepreneurs to focus on these specific industries and solve these challenges with some infrastructural support, the more progress and faster progress we are going to see. 

In the west AI is synonymous with the notion of Skynet and human enslavement when it comes to the general public. But here in India, it rhymes mostly with automation and the fear of job loss. On the other hand, new jobs are created, and new sectors such as data annotating are opening. So how do we address this challenge of reskilling and upskilling in India? 

For skilling, the focus must be on free and accessible programs for every citizen. When I was very young, we were going through this challenge that girls were not getting educated as parents were only sending the boys to schools. To address this, a program was introduced where girls' education was made free up to the 10th standard, which changed the whole narrative. When it's free and easily available, it is much easier for every citizen to be involved. 

I think we have to take a similar approach to reskilling programs, especially on making them more accessible and incentivizing the participants so that more people use them.

Importantly how do we make it fun? It doesn't need to be where you have to go to a classroom and sit and learn. It can be a game on your phone where you are learning about AI. 

If AI literacy becomes part of basic digital literacy, it will change how people perceive AI. Unfortunately, when it's an unknown element, there will be fear. So we should focus on removing those 'unknowns' and educate everybody on what is basic AI. And that's the primary goal of my nonprofit, Humans for AI. 

Humans for AI, the NGO you founded, have been doing some phenomenal work in the space of diversity and in basic AI education. Can you please explain the works Humans for AI is driving?

I've always been an advocate for getting more girls and women into tech technology because often, I am the only woman at the table. And that needs to change. 

Once AI started becoming a reality, and when I was setting up my first data science team, I realized that my first few hires were mostly men. I was the only woman in a team that I was building. That's when I realized I needed to be more aware, and I needed to decide what can I do to address this. 

A lot of what we are doing with AI is encoding and automating human decision making. To some extent, it is about human intelligence and encoding parts of it. So if we don't have that diversity at the team designing or developing these AI systems, then that AI is just going to be incomplete. 

Of course, it will be biased, and it will be unfair. But in general, it is just going to be an incomplete AI because it has not been looked at from different lenses. 

We have seen enough scenarios and headlines where that has happened. So the idea behind, Humans for AI came to me because I realized that diversity is essential, and I didn't see anybody focusing on solving it the way Humans for AI is addressing it right now. 

Anybody who has been part of a data science team or has built data science teams knows that you don't need everybody to be a data scientist on the team to build an AI solution or an AI product. Instead, you need designers, UX, QA, product managers, project managers, subject matter experts and domain experts. 

So, let's not try to teach everybody coding. Let's not ensure everybody learns about AI by learning maths, statistics or machine learning, but surround this homogenous group with diversity. Because then you can pull in the artists, the nurses if you're building a healthcare product; you can build in different skill sets by providing them with a basic AI literacy for them to understand what does machine learning really mean or what does reinforcement learning really mean? We should go beyond those buzzwords to help them understand the meaning so that the basic level of AI literacy can help them leapfrog into a career working with AI products. 

And suppose we focus this literacy program on women and underrepresented minorities. In that case, it's a win-win because now, even when you have a homogenous group of data scientists, he's surrounded by women and people of colour, people from different socio-economic backgrounds. Thus you are making your AI more complete, good, and fairer, but you are also getting diversity in the AI teams. 

Your book "Trustworthy AI: A Business Guide for Navigating Trust and Ethics" in AI will launch on 22nd March. What motivated you to pursue this book, and what are the key messages? 

Currently, there is a lot of discussion on AI ethics, and from my experience working in various industries, ethics can mean different things for different companies. And we have to really solve it and move from these high-level catchy headliners to go down into the weeds.

For example, take fairness. If an AI solution is not working with human data, fairness is probably not as important. Think about if you're using an AI to predict when this laptop will fail and raise a warning in time. Then human data and fairness doesn't really come into the picture. 

But if you're using it to provide patient care or diagnosis, fairness is absolutely important. Let's take one of those controversial ones- facial recognition. We hear a lot of controversy about it, and there are numerous issues with the particular technology. But it all depends on the context where you are using that AI technology. 

The use of facial recognition in a law enforcement scenario or in a justice system is a serious concern, as a lack of fairness can ruin somebody's life. But to unlock your phone or to improve worker safety on a factory floor, it is okay to use facial recognition.

It comes down to that organizational level and use-case level to decide what parts of ethics and trust are important for that particular organization or stakeholders and how you put those guardrails in place. 

The book is about moving from those philosophical discussions to real-world applications within organizations, whether it is a business or a startup or a government. It gives you those tangible tactical ways to think about these different philosophical concepts in today's world.

Of course, regulations are coming. But currently, some level of self-regulation is needed as we are in this interim as no regulations exist. You cannot let AI go rogue everywhere. I think we will figure it out over time. 

Coming to your work at Deloitte, can you tell me about the Global AI Institute and what are projects you are involved in? 

I lead our Deloitte AI Institute, a global institute with different regions involved, and our focus is on applied AI and all of its various dimensions. We mainly look at research and see trends, like where are specific industries heading, what are currently companies doing in this space, so that we can all learn together. And we also go into these fuzzy areas, like ethics, regulations and what kind of governance needs to be in place; what kind of talent you need; how you bring in the right talent, and so on. So, many applied AI aspects are covered under Deloitte AI Institute. 

I mainly focus on trustworthy AI but also on ethical tech. Especially what does ethics mean for emerging technologies? We are all talking about AI ethics, but other technologies are coming at us very rapidly, as we have seen with metaverse, NFT and blockchains. These technologies are all exciting, but there will be negative implications for those as well. And so how do you deal with that so that we avoid what happened with AI, where we are trying to play catch up. 

So those are the areas that I focus on at Deloitte, and it has been very exciting to be at the forefront and to look at all emerging trends, or even at times looking at trends that don't exist, like ethical technology.