The Future of AI in India with Debjani Ghosh
What's the most optimum use of a company's tech investments when it comes to Artificial Intelligence?
With the current hype around AI along with companies scrambling to integrate it into their own processes, what are the real benefits of artificial intelligence when it comes to ROI? What are the opportunities it presents to a country like India? What problems in the AI value chain are primed for businesses to create something cutting-edge? President of Nasscom, Debjani Ghosh and Govindraj Ethiraj explore these burning questions and more in this episode.
TRANSCRIPT
NOTE: This transcript has been done by a machine. Human eyes have gone through it but there might still be errors in some of the text, so please refer to the audio in case you need to clarify any part. If you want to get in touch, do drop us a message on [email protected].
---
Govindraj Ethiraj : Artificial intelligence investments are increasing as companies and chief technology officers look for ways to embed AI into their businesses to become more efficient and competitive. The most visible evidence of this trend is clearly the stock prices of chip companies like Nvidia, which have been shooting through the roof in the last year, meaning that high speed computing AI chips continue to be in demand, as are other aspects of the semiconductor ecosystems. Investments into data centers which house these chips and which will be in many ways, the epicenters of the AI revolution are also increasing in many parts of the world and even in India, but we're already at an AI crossroads. A survey released by KPMG, quoted by Wall Street Journal in July, said that revenue generation has overtaken productivity as the primary gage businesses use to measure AI's return on investment. International Data Corporation, by the way, says that organizations will spend close to $39 billion on generative AI in 2024 alone. A recent Gartner report says that of all the projects being undertaken currently in generative artificial intelligence or genai, around 30% are bound to be dropped by the end of 2025 which is roughly a year's time, going to issues such as poor data quality, inadequate risk controls, escalating costs or unclear business value. The questions are thus quite rightly being posed and at the right time. What's the most optimum use of a company's tech investments at this stage, including in the space of AI? What's the best way to direct these investments? On the other hand, as an IT Services major what is India's role in the future of AI, and where can it effectively plug in and mine both the skills and resources there for opportunities of the future? What are the pathways to the best utilization of India's existing skills to become a significant player in the AI revolution. And it all comes together, the demand and the supply the companies who are going to invest in AI and countries like India, for example, who will provide the skills to do so, answering these questions and focusing our attention on it will also fine tune strategy and effort going forward. I'm now joined by Debjani Ghosh, president of NASSCOM, to address this issue at multiple levels. And first, where are we in the evolution of AI, particularly as seen through the lens of Indian companies, including, of course, startups. And second, the policy pathways and how and why AI needs India. And finally, how should you as a CTO or CEO, be thinking about AI today and tomorrow?
Debjani Ghosh: Open AI came up with chat GPT, November 2022 where I think AI became sort of a household word. It moved from being the domain of engineers to everybody's business, kind of, and everyone started talking about it. So that step change has definitely put a lot of focus on artificial intelligence, lot of investment into artificial intelligence, which is, which is great. But I think, you know, from the launch of chat GPT, from that November till now, we've sort of really been in the hype mode about possibilities and what it can do, and everyone has come out with reports on how it can impact and I don't doubt that. I do believe it is a transformational technology
GE: So artificial Intelligence, as we can see, can clearly enhance many aspects of a company or an enterprise's functioning, for example, enhancing robotics and warehousing. But on the other hand, transportation will still need human drivers. AI will also need human oversight, though it'll allow for handling higher volumes. So we do know what AI can do at this point and how it's lining up to meet the needs of industry and business, but which sectors could benefit more, and how do we look at it?
DG: One is the build and the other is deploying it at the end, customers or clients, be it a bank or a hospital. Etc. Right now. Till now, a lot of the big focus and energy has been restricted to very few companies and few countries, and it has been all about the models, you know, the llms, building the llms, we have billions and billions of dollars because it's compute intensive. It's data and says that the billions of dollars have gone it now as we talk about making it real, I think this is where the opportunity for ID services come into play big time. And what you will see is, let's take a bank, a bank is going to work with a services company, in most cases, to figure out how to leverage the capabilities of generative AI, whether it is the llms or agents etc, to make their own data, to sort of organize their own data and get the insights out of their own data. And I think that's where we are gonna see the monetization capabilities of AI and a lot of investments or lot of focus from services companies, etc. It is about building vertical, domain specific solutions, leveraging the data of the company, organizing that data and then hooking it back to using the capabilities of AI or generative AI. So, you know, in my mind, it's not just a generator. AI plays end to end, because data analytics come into play. Machine learning comes into play, and now all of the security comes into play. It's sort of your end to end AI stack that has to get built out. So that's where I see the big opportunity for services and for customers. Again, it's about, you know, a lot of companies have told me that they were they're building their AI strategies, and I absolutely get very scared when I hear that, because I don't know what an AI strategy is. What is your business strategy, and within that, what problems is AI Best Place to solve? You know, that's how CEOs have to start thinking, or CDOs, or I think, I think it has to start with CEOs, but you cannot have a separate AI strategy and Dari. One of the reasons why we are not seeing big impact is because you have companies creating this little vertical thingy called, you know, your AI team or AI strategy. It has to get integrated into your business process. So I think we are beginning to see that trend. The BFSI sector is definitely leading it. We're definitely seeing a little bit more traction in retail, tremendous opportunity in manufacturing, tremendous opportunity in logistics, supply chain management, but we will see much more traction happening over the years.
DG: So the journey to becoming an AI first company has just started, so I think it's very early for course correction, but it's really a revolution. You started off getting excited with the generative AI capabilities, and then you started realizing that what you really need is sort of an end to end thought process in terms of, how do you build an intelligent organization or intelligent processes? I think it's an evolution that's happening as companies are playing with it. They're beginning to understand more and more in terms of how it can be used, how it can be deployed. We still have miles to go open, because even if you look at what is happening today, most of the deployments you are seeing in the space of conversational AI chat bots, you know, these kind of things coming customer engagement tools, etc, etc, that is coming into play. There are very few companies that is really looking at internal data organization, end to end building end to end analytics capacity. That's where I believe the real ROI sits, and that's where also, if you think of it, from companies that are building that's where the monetizing opportunity sets. But that's an evolution, and it is slowly. I mean, I was very impressed with the Bank of Baroda. They actually have worked, if I'm not wrong with Accenture, and they have completely changed. I mean, I think they have built one of the largest data lakes with their own data see if companies just want to use the lnms as is, be it GPT llama or whatever have you, you only had to get so much returns from the investment you are making and and there are problems of bias and hallucination and all of that coming in. The magic happens. You bring your own data. The magic happens when you break the data silos in your organization, and you build the capabilities end to end, to organize that data, to get the analytics out of their data.
GE: Artificial Intelligence is a hungry beast and has to be fed with data all the time, and there is much more to data itself. For example, AI in driverless cars face challenges linked to human intuition. Now there's a lot of AI training going on to help algorithms identify via the camera lens, obviously, objects like trees, pedestrians and other vehicles. Now this is all physical effort. Incidentally, of course, it's an opportunity too, but let's understand what data exactly means and how it breaks up into multiple aspects. And let's hear what Nitin said, CEO of Incedo, a company that works in AI solutions, and author of the book The data paradox, and also former chairman of NASSCOM, GCC Council, has to say
Nitin Seth: There are four type of data problems. You know, the first problem is talked about it, which is, but I'll put it differently. It's got contextualizing the data. And I would some it's not about either or Govan. It's not saying that, Oh, you know, it's all Gen AI and the data of the world, or it's enterprise data within you need both. Let's be clear about it. Yeah, so you need both. But then you know, how do you kind of, you know, what do you need? Yeah, of that, what do you need? What do you use for? Second is the data. Once you are bringing different data together outside, inside, multiple internal sources, the second problem is the data integration. How do you bring it together? Because these are very, fundamentally, very different. These are not even apples to oranges. This is like, completely different planets. Yeah. So how do you bring that, you know? How do you integrate that data? You can't do it physically. How do you do it logically? Then the third question, once you kind of, you know, progress on data integration, is the data quality question that you know, you know garbage in, garbage out, we always know that. Now the challenge that is happening covid And start is that the it's not just about the traditional measures of data quality, about accuracy and timeliness and blah, blah, blah. It is also about context. In a particular context, a particular type of data may be okay. In other context, it is different. So the data quality question is becoming more complex. And the fourth is about data security, which is at this point, and especially if you're in a regulated industry like, you know, like financial services, you know it is, or in healthcare, they have a slightly different concern, which is about personal data. Data Security is a huge issue. And you know, how does your data flow work? Because, you know, the the nature of AI is that's a bi directional flow that, you know, you're getting the learning benefit, but what are you contributing back to it? You know, how do you protect that? So those are the four issues, you know, when you break down the high level of, you know, the bias kind of type of question which makes all the headlines and enterprises are aware of it? Yeah, this is how it breaks down, that what data you bring together and contextualize it, how you integrate it, how you solve for data quality. And the fourth is data security.
GE: So the larger question could be, what's the AI value chain like, which end of the spectrum lies the biggest opportunity? And of course, what makes the most sense for a country like India, and let's hear what the Debjani Ghosh has to say.
DG: What's the AI value chain? So it on top of the value chain is the infrastructure. Those who control the infrastructure will supposedly control the biggest share of the pie, right? An infrastructure is cloud, data, compute, etc, etc, right? Then you have the next level, which is where you may not have a control on the infrastructure, but you do control a lot of the vertical build out that's happening. So you control the end customer data, which you're bringing in, which, again, becomes very, very important, right? The third phase is where you are using voting says to build your applications. They're generally known as wrappers, where you're building your wrappers, etc. But that becomes difficult to defend. It's all about who builds it better, who builds it faster and who builds it cheaper, right? So that becomes a little difficult to defend. So I think today, if you look at what's happening, especially at the startup space, a lot of it is in the early phase. And I think what is needed for India is to move up towards the middle, where we really become the preferred partner for all verticals, for creating those vertical solutions, as well as Moats. I think that's where we really have to play, both startups as well as companies. I honestly think India should not give up the fight or give up the journey. I won't call it a race to build some level of infrastructure, because we can, and if we don't. Do it in a few years, we are going to have the same crib that we have today, that we didn't invest early enough or enough in the cloud in building a hyperscaler out of India. And therefore today all our data is going out. I think you're going to have the same issue. So I do believe that we have to move up the value chain. Yes, a lot of focus from industry will be on moving to the next level, which is building the vertical solutions and more. But I do believe government, especially and academia, has a role to play in building some of the infrastructure level, especially with respect to compute, because otherwise, while data can be a great mode, because a competitive advantage, but if you don't have the compute, you're not going to be able to do a lot with it from building a competitive advantage perspective.
GE: So let's take the big picture and the opportunities and the role for India in the AI race, if you want to call it that. And here's what Dr Alok Agarwal, author of the book The Fourth Industrial Revolution and 100 years of AI, and founder and CEO of Scry.ai he also was earlier founder and director of IBM Research Labs, and how he sees the opportunity
Dr Alok Aggarwal: The market will be huge for India. Data annotation is a very new category and helps. It's not very, very specifically defined, broadly different. So it contains all kinds of people who are doing data annotation and groundwork technique. For example, if a medical transcriptionist who was transcribing medical records until yesterday, could be now a data annotator, because it's now the AI system which is transcribing, but someone has to check how correct it is and and if it's not correct, it's so currently, according to US, there are about 400,000 data annotators in the world, and about 200,000 are in India. About 100,000 are in China, and remaining 100,000 in the rest of the world, including Philippines, which has a large contingent and ASIC and because of the low wage countries, I think that will continue to face. Out of 15 million, we are expecting about 9 million to come from India, because India has very good proficiency in English. So it's almost like a next wave of BPO, isn't it? It is. There is some consolation India should have, because part of the jobs will be lost. I mean, if we look at two decades from now in call center domain, because a lot of the work that all center agents are doing will be done at least the recommendations will be provided by llms and by AI systems. Now I may still be at all centered in reviewing those recommendations and then giving it to you the customer, but it would take away the likelihood of my doing all the work, and therefore the number of agents will reduce. Similar number of medical transcriptionists will reduce, because AI will be doing the medical transcription. But on the other hand, the ground truth checking within so in some sense, there would be a trade off. I think overall, India will gain more because we have only 1.5 million call center operators or agents right now and a number of medical transcriptions. So overall, India will bring quite a bit. And I think this is a good thing for India. Another good thing is that these people can be easily located in second year, third year, cities. So it's good for the cities also.
GE: So that brings us back to the big opportunity and question that we posed at the beginning, how important is India's vast pool of skills and expertise in this emerging AI world, even as it recalibrates to business needs and objectives? Or is the question the other way around? And let's hear what the Debjnai Ghosh president of nascom has to say,
DG: I do believe that AI needs India, because one of the biggest problems with AI. I mean, given the kind of investment that has gone in returns, I mean, you are you're already hearing you mentioned it, I mentioned it, the voices questioning returns is gonna get louder, and justifiably so, right? This is where the world has to start thinking about to get the kind of returns that is needed to justify the investment that's going into building AI. We have to drive population scale deployment. That's the only way you can get these kind of returns. I think India is one country that has figured out how to practice
