‘Courts and law enforcement need to build AI capabilities’

Statistically, what is India’s share in Microsoft’s operations?

No one should have any doubt about the importance of India—whether it is about Microsoft or other tech companies. It is one of the fastest-growing markets for Microsoft globally. That is a reflection of this enormous spurt in digital growth India has been on for the last five years—it has accelerated and shows no sign of abating. It’s good for India, and certainly good for us. That said, we don’t give a geography-wise breakdown of our numbers.

What kind of AI regulations are needed across countries and what would Microsoft want from India?

There are many existing laws and regulations that we should all remember will be applicable. For instance, if it’s unlawful to defraud someone, then it’s also unlawful to use AI to do that. No one can go before a judge and blame a technology. First is to ensure courts and law enforcement agencies develop AI skills and capabilities to evaluate it. It also means that organizations subject to these laws must develop capability to use AI properly. That’s a broad goal, and is a critical starting point.

Then comes AI foundation or frontier models. That’s where things will get a little more challenging from the perspective that people will have to decide what kind of regulation they want.

We must put safety as a very high priority, and one of the things it needs is proper testing when the models are developed. Standards are emerging with the use of AI Red Teams—independent teams evaluating the models to identify risks, and state whether the steps taken to mitigate the risks are successful.

There must also be some level of global standardization or reciprocity (of regulation). I don’t think one would be well served to have every country in the G20 asking a Red Team to work differently. There’s a role that national activity would play, but with global coordination.

Microsoft will have to deal with a host of new and upcoming regulations that are still evolving. How are you preparing for it?

Life is understandably getting more complicated for the tech sector. On balance, it’s probably a good thing. It is part of the maturation of the tech sector. We’ve been working on it. Two years ago, we launched an initiative focusing on what we call regulatory governance. We asked what will be most applicable to our kinds of technology? For most areas, we have ended looking at financial services, which are very well and carefully regulated but allow ongoing innovation and growth.

Similarly, we could have know-your-cloud obligation, to ensure these advanced applications and models are running secure cloud environments. There could be a know-your-content regulation. People will really know the outcome when all regulations are in place. However, the fundamental formula India has adopted for privacy is sound. The focus is on the role of data fiduciaries, in terms of their responsibility, and the role of consent.

A question being asked increasingly is if there’s regulation of AI, then where should it be done?

At an applications level, I would say do it where it’s already done. In other words, if it’s financial services or healthcare, you look to those sectors’ regulators for these models. We’re more positive about the notion of having a specialized regulator—aircraft are regulated by an aviation regulator, and automobiles are regulated by an automobile regulator. It’s all based on understanding the specifics of the technology. So, I don’t think it’s implausible or radical to say that a new technology might just be best governed by a new regulator for that technology.

The answer will vary in different countries, but having that conversation is useful.

In a country like India or elsewhere in the global south, one tends to see IT ministries that are closer to being able to be that technology regulator, than what you see in Europe or North America. This is where I feel that governments now tend to leapfrog older governments.

What are the concerns of generative AI being weaponized and used in cyber warfare, and what can Microsoft do about it?

I think there’s a lot that we can do. There are areas where criminals undoubtedly turned to AI, like any technology, to create stronger weapons. I’m probably more concerned in the area of deepfakes or cyber influence operations. Yet, the more work we do in this space, the more I fundamentally believe that if we invest with ambition, we can use this technology to build stronger defences at a pace that is faster than people can weaponize it.

Take disinformation that one has to worry about in the context and political campaigns—AI can enable people to move faster or produce something better, but the truth is they can produce a lot today. I think what we really need to do is embrace the opportunity to harness the power of AI to identify patterns and data, because we’re awash with data.

A lot unfortunately happens in plain sight today, so the ability to sift through that data, for us to find things faster to alert the public, to combat it—I think that is going to be a bigger game-changer than the advances in weaponry.

Is it a concern that we do not have enough skilled resources in cyber security?

It’s definitely a challenge, and that is why we’re investing so much in cyber security skilling programs, including in India. We should be able to catch up (with the shortage), because these are very good jobs—they tend to be well-paid, and we need to double-down on that.

But wherever you have a labour shortage, AI probably has more potential than in other areas, because it’s such a productivity enhancer. This is all the more reason for us to lean into the opportunity to use AI to strengthen our ability to protect against something like cyber security threats.

Does Microsoft have a specific plan to build the capacity of skilled cyber security professionals, even in India?

We have specific plans, yes, and in fact programs that we have been implementing here in India. One of the things I was really encouraged by when I was here a year ago was to meet some of the people in our cyber security skilling programs. In particular, the meeting that I remember the most in terms of being the most inspiring was to see young women who had specifically enrolled from tier-II and tier-III cities. They had quickly mastered the cyber security skills needed.

Part of the solution is to do a better job, as an industry, of recruiting women and making them feel very welcome. That’s going to be true for cyber security, and it is going to be true for AI as well. We need to partner with more non-profits, and encourage the government to increase its investment level.

This is one problem that we can advance and address more quickly by bringing in more resources.

There is a lot of fragmentation in the way tech develops. When we look at generative AI, is there a possibility for companies to come together to collaborate and build unified standards to address the collective issues?

I think there is. There’s a Frontier Model Forum (announced by Microsoft, Google, OpenAI and others on 26 July) that we’ve created with a small number of companies, and it will grow. We need to recognize that there’s a common approach to safety standards that we really need. If you look at us and Google—we compete vigorously every day, and yet we have come together.

That’s the type of thing we need to encourage. It also helps when people in governments bring us together, because it’s not a natural act. This is where the White House in the US has played a critical role in bringing us together, and we’ll probably benefit if the government leaders, in particular capitals, continue to do that.

Does it worry Microsoft that some of the common narrative on AI is being overhyped and the promises of that may not be that realistic? Also, taking Azure and its success into account, how important is India as a market for Microsoft?

India is an extremely important market for Microsoft, and inevitably it will become even more important every single year. Ultimately, we live in a world where GDP will correlate with population. As countries grow economically and we see economic growth become more broadly distributed around the world, the single best predictor of GDP a decade from now is population. India has already made the leap to turn that into reality.

While India’s digital transformation started nearly two decades ago, in this decade, India has digitally accelerated at a completely different pace. I do think that AI is a game-changer, but I also think that there has been a hype cycle, and AI is not necessarily something that serves everybody. Look at the past six months—around the world, we went from looking at generative AI as something that actually works to something that will save the world to something that will destroy us all.

The truth is, it’s not going to save humanity, or destroy humanity. It is a tool, one that we need to learn how to use, and invest in improving. It is a tool that can play a profoundly impactful role over time—in providing what could be the greatest technological advance of our lifetime. It can help us pursue new drug discovery, improving education and really advancing productivity. But, it’s not going to replace the need for people to think.

People who learn how to use it will think better. It definitely can enhance our capacity to create, think critically, express ourselves and analyse data. It has enormous potential, but there is a hype cycle.

What is your perception of how India has changed over the past five to seven years?

There’s no country in the world that moved as rapidly during the pandemic to spread the adoption of digital technology. This is obviously represented in the digital public infrastructure (DPI) you have here. You do not see anything like that scale anywhere else in the world. To some degree, that’s not surprising given the size of the population, but even if you just do it as a percentage of the population and see how many people it has reached, it is extraordinary.

Adding to that is the vibrancy of the startup community in India—what we call ‘digital natives’, or companies born in the cloud. That has exploded. I’m very struck by the degree to which innovation and technical skilling is reaching into the tier-II and tier-III cities. You add those advances in digital technology with demographic factors that are unique to India—one of every four people of working age, between the ages of 20 and 64, lives here.

Maybe more than any other country on planet Earth, the influence of India through its people doesn’t stop at its border. There is a massive number of talented people from India who live around the world, who have taken engineering or skills they first acquired here, and are now connected not just back with India but with each other globally.

What really explains why Indians have been able to adopt the digital journey, with the entire India Stack, faster and better than any other nation?

Part of the explanation is the very large number of people who have gotten an engineering background—nearly half a million of them are being added in India every year. Whenever you have a population with that kind of foundation, it helps you spread technology adoption more broadly.

But there’s a second aspect—around the world, you’ll not only encounter Indians who are local successes, but they also tend to be involved in their communities, and civically. If you go to the Midwest in the United States, in smaller cities or towns, if there is a community effort to raise money for a new hospital, you’re highly likely to find someone from India—who has succeeded in the local business community—playing a role. There’s clearly something in the ethos here that encourages civic-mindedness. In some ways, that’s bigger than the technology phenomenon.

Updated: 28 Aug 2023, 07:00 AM IST