Prime Venture Partners Podcast
A podcast for entrepreneurs who are looking to build & grow their startups. Avoid common traps & learn uncommon strategies & tactics from makers & doers of startup ecosystem. Prime Ventures is a early-stage venture fund which focuses on startups that not only need capital but also require mentoring to transform them into disruptive companies. We share a passion for working closely with entrepreneurs and enjoy sharing their journey in a high-frequency, interactive and fun environment.Read more about us at http://primevp.in
Prime Venture Partners Podcast
The Truth about Artificial Intelligence (AI) with Dr.Alok Aggarwal (Author, Scientist, Entrepreneur)
In this episode, we hosted a globally renowned and prolific guest Dr. Alok Aggarwal, founder, CEO, and Chief Data Scientist of Scry AI, the Author of the book "The Fourth Industrial Revolution & 100 Years of AI (1950-2050)" and an Inventor with 8 patents.
Dr. Aggarwal pioneered the concept of “Knowledge Process Outsourcing (KPO)”, “co-founded” Evalueserve (4000+, employees), “founded” IBM’s India Research Laboratory, founded Scry AI that builds proprietary AI products for enterprises globally.
He has published 125 research articles, taught 2 courses at the Massachusetts Institute of Technology (MIT), has a Ph.D from Johns Hopkins University and a B. Tech. from the Indian Institute of Technology (IIT) Delhi.
In this conversation with Pankaj, with insights drawn from his book, "Fourth Industrial Revolution in 100 Years of AI from 1950 to 2050," Dr. Alok presents a compelling argument for why AI is not just another technological trend but a catalyst for a new industrial revolution. He delves into the history of industrial revolutions to understand what makes AI stand out.
From steam engines to CPUs, each era has been marked by inventions that transformed societies. This episode offers a thorough analysis of how AI compares to these past innovations, while also cautioning against the hype that surrounds it.
He explains how AI's unique capabilities in classification, pattern recognition, and data processing are reshaping industries from banking and technology to healthcare and heavy engineering.
For entrepreneurs, the episode highlights the risks of getting caught up in AI hype without developing robust intellectual property and suggests strategies for creating high-value AI products.
In this podcast episode we spoke about the below topics, dive in:
03:55 - Historical Analysis of Industrial Revolutions
19:11 - The Impact of AI on Industries
34:05 - Navigating AI and Intellectual Property
45:27 - AI Transforming Services in India
Enjoyed the podcast? Please consider leaving a review on Apple Podcasts and subscribe wherever you are listening to this.
Follow Prime Venture Partners:
LinkedIn: https://www.linkedin.com/company/primevp/
Twitter: https://twitter.com/Primevp_in
This podcast is for you. Do let us know what you like about the podcast, what you don't like, the guests you'd like to have on the podcast and the topics you'd like us to cover in future episodes.
Please share your feedback here: https://primevp.in/podcastfeedback
Unfortunately, ai is again being hyped very, very strong. Businessweek, one of the magazines in the US, actually had the front cover in 1984, I believe it says AI. It's finally here. People get all excited and create hype cycles. Ai machines are essentially chai machines. You train them just like you train your children to provide learning. Then you test them. 7-8% of the US economy easily runs on COBOL and most of the US Army Department of Defense runs on COBOL. So some of the nuclear submarines run on COBOL.
Pankaj Agrawal:The GPT-40 is like a Ferrari for grocery shopping.
Dr.Alok Aggarwal:Mckinsey says it's about $15 trillion will come out of it by 2030. Everyone talks about China and US. People are forgetting that India has the largest number of software professionals right now in the world. Metaverse probably will occur more like 15 to 20 years from now. Ai is not machine learning and machine learning is not AI.
Pankaj Agrawal:Hello everyone, my pleasure to welcome all of you for the new episode of Prime Podcast. The guest I have today is someone that I've been really looking forward to chat with. He's an engineer and computer scientist. He spent 15 plus years with IBM and did deep research at the intersection of computer science operations and data scientist. He spent 15 plus years with IBM and did deep research at the intersection of computer science operations and data analytics. He holds nine patents and several research publication. He was also the founding director of IBM research in India office, which he helped set up at IT Delhi, his alma mater.
Pankaj Agrawal:He later went on to found E-ValueServe, a professional services company that provides research in analytics, services in intellectual property, legal processes, market and business research and data analytics. It went on to become a global company with offices and research centers in 15-plus countries. In his most recent role, he's contributing meaningfully in AI by founding Scry AI, which provides AI-based product solutions and services in BFSI technology and heavy engineering and life sciences and healthcare industries. He's currently running the company as the CEO. To top all of that, he recently authored a book Demystifying AI called Fourth Industrial Revolution in 100 Years of AI from 1950 to 2050.
Pankaj Agrawal:Dr Alok, it's my pleasure to welcome you and I'm glad that we are doing this conversation. Thank you for having me Great. So to kick this conversation off right off the bat, dr Alok, I'm reading your book and what I found very interesting is that you mentioned that AI has brought about the fourth industrial revolution and with the past three, kind of triggered by the discovery of steam engines, electrical motor and, most importantly, cpus. What are the characteristics of the industrial revolution and why do you think AI has the potential or is actually in the midst of drawing out a fourth one? What is so fundamentally different about AI than the previous revolutions?
Dr.Alok Aggarwal:So every revolution, industrial revolution, has its distinctness. The interesting part, when I started writing the book is I did not realize that nobody had really defined what an industrial revolution was. So actually, the first chapter sets up a framework for industrial revolutions and what are their characteristics. How do we know that this is the fourth revolution, not the fifth one or the third one, and so on. So from that perspective, the first chapter actually sets up eight characteristics of all industrial revolutions that I have seen, and I'm not saying by any means that these are all the characteristics. There may be more that I have not discussed, or I do not even know about.
Dr.Alok Aggarwal:The first one, typically in each industrial revolution has been that it wasn't formed by one invention. It was formed by several inventions and they actually often used each other to improve. So an inventor or inventors would use one invention to improve another and so on. So the first, for example, as you said, the main invention was the steam engine, which was being used almost throughout, and it became pervasive and very well known. So water and steam infrastructure was created, and what steam was obviously needed for steam engines? Steam also started as an infrastructure. Water and steam also started fueling flower mills, for example converting wheat to flour, textile mills and so on. So the first characteristic of all these inventions is that there are several inventions. I'll come to other revolutions in a minute. The second is that it at least creates one infrastructure for the society. In the first revolution it was water and steam infrastructure. In the second one it was with because of electricity or electric motors. Electric generation and dissemination or distribution was the infrastructure and interestingly, electricity not only propelled or fueled electric motors but also electric bulbs and many other things which have nothing to do with motors. The third infrastructure again had several inventions, but the one which became the infrastructure going forward was electronic communication, wireless Wi-Fi and so on. Before that, 1950, I mean 1995, most of us were using dialing, using basically phone lines to dial from home to the office, which was extremely slow, and that basically converted, fundamentally changed that. So those are the first three. And the third characteristic is that there is one invention which basically becomes pervasive during the entire period and it becomes ubiquitous. For example, steam engines began to be used pretty much everywhere in the 1800s, all the way from steam cranes or railroads to steamboats, steamships and so on.
Dr.Alok Aggarwal:Second one electric motors. There are about 3,500 different types of electric motors. They're being used everywhere. In my room sitting here, I'm sure there are at least six or eight electric motors running in the air fan, in the air conditioning unit and so on. And then there are five other characteristics which we can go through in more detail.
Dr.Alok Aggarwal:But, going back to your question, this particular industrial revolution has many inventions, so not only AI and inventions related to data, but also inventions related to climate change, related to Internet of Things, robotics, gene editing, personalized healthcare, blockchain, metaverse, ar, vr. Not that all of these will happen immediately. In fact, most of these revolutions were there for 40 to 80 years. So we started here in 2011 for the fourth industrial revolution. Probably it will continue till 2050, if not 2060. And some of these inventions, like Metaverse, probably will occur more like 15 to 20 years from now, because we don't even have the high-order broadband communication or 60 wireless communication. But we can discuss the other five as we go along.
Dr.Alok Aggarwal:Why AI is different is that each of these inventions, in each revolution which became pervasive, was markedly different than the previous one. For example, steam generators or steam engines were working very differently than motors, and it moved from effectively what I would call a resource-based infrastructure, which is the resource was water or steam to an electricity-based infrastructure which was created by humans. Electricity was being generated by humans. The third infrastructure was quite different from the second because the infrastructure was that in communication and there was nothing being generated, so to speak. It was just being communicated from one end to the other.
Dr.Alok Aggarwal:And the fourth infrastructure, with AI, it's now about data, which we can say it's very much like electricity or like steam engines I mean steam or water or like broadband but data has many, many facets and it's very much different. So, although people say data is the new oil or like broadband, but data has many, many facets and it's very much different. So, although people say data is the new oil or data is the new electricity, they're actually obfuscating a lot of the issues or a lot of the important points that data has. And AI is, of course, using data to be the next pervasive invention. We already have people do not realize often already have more than 2,500 use cases of AI which are operational in nature. Everyone is today's hyped up about LLMs and GPTs, but actually there many, many instances where AI is already being used and will be used by 2050, my belief is at least in 100,000 use cases.
Pankaj Agrawal:AI by design, what you can say a lot more intelligent than the previous three. It has a lot more capabilities around reasoning, about contextual understanding, about vision understanding and basis, which it can enable, as you said, right, multiple of these use cases. Right, so do you think, um, because of that and you know, for example, many of the big tech leaders whether it's on that pitch I came out with the statement, right, you know that invention of ai is is as, uh, as instrumental, if not more, than electricity, than the invention of electricity. Right Now, that's a big statement. Arguably, that there are vested interests and so on and so forth, right, but do you kind of believe that the acceleration or the unlock of use cases through AI could be a lot more pervasive than what the previous ones have been? Or do you think, you know, it's too early to kind of really figure that out, because AI is fundamentally, you know, I would say, meaningful intelligence over the past three, right, so do you think it would affect how it gets kind of seeped into the mainstream?
Dr.Alok Aggarwal:Yes, two aspects. One is that, unfortunately, ai is again being hyped very, very strongly, partly by various technology companies, partly by the media. This is not the first time that this hype has happened, in fact, when artificial intelligence was initially created by Alan Turing by talking about the imitation game. After that there was a conference in 1956 in Dartmouth College, and that's when the name was given. And during 1956 to 1973, ai was hyped equally strongly, maybe more, maybe a bit less, but about the same in many ways. In fact, a movie was created, 2001, a Space Odyssey, in which the computer HAL 9000, is considered to be an artificial general intelligence computer with emotions, abilities to scheme, arrogance, all the human characteristics and much more, and it can beat people in chess and so on. So, from that perspective, that was the first hype. It went sour, it went bust in 1973. The second hype started in 1980 with expert systems, and that's a very important thing. What is AI? Ai, in effect, is a combination of machine learning and expert systems, expert systems being knowledge-based, subject matter expertise. And even for humans, without subject matter expertise, we are not. Without subject matter expertise, we are not able to do most of the work. We become eventually knowledge workers in some field or the other or are doing a basic work like typing and so on. So, from that perspective, 1980s again there was a. It was short-lived, it went bust in 1987, 88. And then this is now again the hype. Interestingly, all hype cycles have the same thing. Businessweek, one of the magazines in US, actually had the front cover in 1984, I believe. It says AI it's finally here. And isn't that almost what everyone is saying, including Mr Pichai and others AI is finally here.
Dr.Alok Aggarwal:Having said that, all inventions that is the fourth characteristic I bring out of industrial revolutions All industrial revolutions have these very important inventions, but they take time to seep into the society. Things did not happen overnight. For example, railroads. The first railroad was created in 1830 between Liverpool and Manchester and people got so excited. Obviously steam engines were getting pervasive. People got so excited. Inventors, investors and then the public, media, public they all put in a lot of money. By 1890, it was oversaturated, the hype had died down. Most of the railroads went bust. Same thing we see during the second industrial revolution with telegraphs. In the third industrial revolution with telecoms and dot-coms I mean the telecom money. About $500 billion was spent in billion, not million. $500 billion was spent in to lay cables in sea, under water, that is, under soil, wireless Wi-Fi and many of these companies went bust.
Dr.Alok Aggarwal:So it takes time for inventions to seep into the society and of course, the time is reducing but it is not going away. For example, in 1882, the first electric generation plant was created by Thomas Elber Edison in upstate New York. That was in 1882, but it was only in 1925, that is, about 43 years later, that half of Americans got electricity. So 43 years later that half of Americans got electricity. So 43 years later. Similarly, even more recently, the first handheld phone was created in Motorola by Cooper in 1971. But it was only in the 1990s that people actually started using handheld phones. I'm just talking about not even smartphones like BlackBerry. Yeah, started using handheld phones. I'm just talking about not even smartphones like Blackberry, or that is something which is very important to keep in mind that things take time to seep into society.
Dr.Alok Aggarwal:I give eight reasons why that's the case. Everyone falls into that, including me. When I started EvaluServe, we gave the notion of knowledge process outsourcing and we said in 11 years or 10 years, this is how much it will be and India will gain so much out of it. The rest of the world will gain so much out of it. Not that fundamentally we were wrong. We were wrong about the time. It took at least two times what we had projected, what I had projected.
Dr.Alok Aggarwal:And that is the fifth characteristic of industrial revolutions. Fourth being things take time to seep into the society. Fifth being that people get all excited and create hype cycles which go boom in the bus. Of course, many of these hype cycles are good. I mean, if there was no hype cycle, we would probably have no railroads. Most of the railroads that we have to do are because of the hype cycle that got created in England and in the US, and that's a good thing. So I actually am a very strong believer. It may hurt the investors, who may lose their shirt, but it helps the society in the long run.
Dr.Alok Aggarwal:So definitely, ai will play a big role, probably a bigger role than electricity, although that's hard to say at this point, but I think that will happen. Gradually the pace will begin to pick up, but it will happen. As I said, we clearly have about 2,500 use cases that I know of and I'm putting up about 1,000 of them by the end of August on our website. There are many more and I think it will go to 100,000. So, from that perspective, there are only 3,500 motors electric motors of different kinds but there may be about 100,000 different use cases or different type of AI systems. So from that perspective, mr Pichai may be right.
Pankaj Agrawal:Excellent. I think that's a good segue to discuss. What is the present state of AI right? You mentioned about 2,500 use cases out there and, with your experience at SCRI and the research for the book, what sectors and industries are you most excited about with respect to AI adoption? Maybe some examples of the use cases that are getting accelerated or maybe altogether replaced by AI first approach, if you want to pick your favorite out of those thousand.
Dr.Alok Aggarwal:So, first of all, I think pretty much all of the industries and departments of organizations, or you can say divisions of organizations, will be affected by by here. There is absolutely no doubt in my mind that, if we excuse me if we look at the long trajectory I mean if we assume that the revolution started in fourth industrial revolution started in 2011, with IBM winning the Jeopardy contest my feeling is we have at least another perhaps 10 to 15 more years to go before AI really seeps deeply into it. So, as part of those 1,000 use cases, pretty much I would say, are broken up into 19 industries, one government with 20, and then 10 different departments or different divisions of a typical organization finance and accounting, procurement, marketing and sales, and so on and so forth. So, from a perspective of companies, you can look at AI. Ultimately, if we take on the hype, what can AI do today or in the near future? One thing it is very good at is classifying, that is, it differentiates between the faces of dogs and cats. That's a very simple example in vision. But a more interesting example is to differentiate whether a person has skin cancer or a person should be given out a loan. So these are classification. Secondly, it is very good at figuring out patterns. So otherwise, AI, as Alan Turing said, AI machines are essentially child machines. You train them, just like you train your children to provide learning. Then you test them and if the test fails it doesn't have enough accuracy to train more, or you rewrite the AI program.
Dr.Alok Aggarwal:So from that perspective, already there is a lot of innovation, a lot of use cases in the Internet of Things, For example, figuring out whether somebody has a water leakage or a gas leakage in his or her house. There is about 4% water around the world gets leaked in water pipes as the water companies send the purified water, the potable water, from their plants to homes. So a lot of this is getting now, because the sensors are relatively inexpensive, that a lot of this is happening already and will continue to happen in the area of supply chain management and in general internet, happen in the area of supply chain management and in general internet. Clearly that would be one of the big fields, if I would call a vertical which will get affected. Another one will be banking and finance insurance. Now, insurance has a very interesting problem, especially life insurance, because it goes on for 40, 50 years and therefore they have so much data which is in paper-based data and that paper is not the facts of today very, very poor quality. So converting no matter. Even humans cannot understand it, Forget about AI understanding it. So that's a very important problem and I think it will be resolved in the next 10 to 12 years.
Dr.Alok Aggarwal:Another problem which is along the same lines is reverse engineering of COBOL programs. So COBOL was a language which was created in 1960-61 and even though it's a very simple language, because it's a simple language, people never wrote documentation on it. And even though it's a very simple language, because it's a simple language, people never wrote documentation on it. And even today about 7 to 8 percent of US economy easily runs on COBOL. So all of insurance companies today I mean India fortunately got saved because COBOL by 1980s had become already an older language and newer languages had come in Java and so on C++, C, Java.
Dr.Alok Aggarwal:But COBOL for the Western world, for the more developed world, is a very, very big problem because people like me I mean I learned COBOL in IT in 1975, one course when, when we did not have even tapes, we had a deck of cards which we used to feed into the computers and I mean people like me pretty much. Most of them have retired, Some even unfortunately passed away. There is no documentation and what happens all the time is that people modify their COBOL programs and those have bugs in them. Nobody can now figure out where are the bugs. So just to give you an idea, most of the US Army Department of Defense runs on COBOL.
Dr.Alok Aggarwal:So some of the nuclear basically reverse engineering COBOL programs so that you don't need effectively, you need very few if any, COBOL programmers. The other advantage of reverse engineering of COBOL program using AI is because it doesn't know the syntax of COBOL programmers. The other advantage of reverse engineering of COBOL program using AI is because it doesn't know the syntax of COBOL. It would actually find all the errors, or most of the errors that are in the COBOL program, because it's not transliterating COBOL program into a Python or a Java program but into a flow chart.
Dr.Alok Aggarwal:So then we've got many cases which are very interesting. I mean, healthcare is another one which will be fundamentally changed in the next 10 to 14 years. But again the hype far exceeds what reality is, which is the sad part. And right now I mean, on the other hand, as I write in the book, hype is good, because if we put in 50 billion dollars in in llms and gpts, it is good for the humanity, the human society, because something good would come out. Problems are hard problems, they're not easy. Yeah, so you need a lot of people in a lot of months.
Pankaj Agrawal:Interesting, and I think that's a good segue to the next set of topics that I had in mind. I mean, given that you're serving large industries of all shapes and sizes first with EVL, you serve now with Scry where are these companies in the AI journey? Because what we constantly hear is that a lot of pilot testing is going on but barely anything has moved into production. Sequoia came out with a recent report that a lot of CapEx is being spent in enabling generative AI and powering LLMs. Kind of make up for that, right.
Pankaj Agrawal:Google, meta, microsoft, all these guys I think they recently had their quarterly earnings and all of them are spending anywhere from 15 to $20 billion every year, every quarter, not every year every quarter as CapEx. A large chunk of that is obviously going into procuring the GPUs, you know, obviously. And then NVIDIA is obviously NVIDIA stock has been, you know, touching the skies, but none of that has resulted into the kind of revenue, right? Or maybe the production use cases, right? So what do you hear from your customers, right? I mean, where are they in the journey, right Among?
Pankaj Agrawal:the use cases which are kind of already proven. Yeah.
Dr.Alok Aggarwal:Yeah. So again I go back to the characteristics of the Industrial Revolution. I said the fourth characteristic was that it takes time for even the best inventions to seep into the society, the fifth being that people get hyped up about it and therefore you have a boom-bust cycle. And we are in a very strong boom-bust cycle on Google Buzz site, because the problem is, everyone has gotten excited about one very, very specific aspect of AI, which is GPTs and LLMs, large language models and generative pre-trained transform. The first paper in this regard came from Google Research itself. It was a research paper called Attention is All you Need, and after that Google created BERT, which was the first transformer. It wasn't that good because it didn't have enough parameters, so to speak, without going into details, but since then the race, or the war, began.
Dr.Alok Aggarwal:Llms are very interesting for humans because it can write me a recommendation letter, it can summarize very large pieces of text for me, it can improve my English. They've been trained. These particular deep learning models, or deep learning networks, are trained on anywhere from 400 million pages to a billion pages of text and tables and so on. So they're very, very good in writing English. They're almost perfect in that. Imagine a kid going through that many novels If a novel is about 400 pages long that's a large novel, maybe 200 pages long, talking about 2 million novels Then either the kid will go crazy, but also the kid will probably become excellent in writing English, very cogent, very proficient, and that's what you see in LLMs also. They're extremely cogent, they're extremely proficient, they're so good that they can even fool lawyers and they fool lawyers into believing that non-existing cases exist and therefore lawyers into believing that non-existing cases exist. So, just like I was talking about the kid, that the kid will go crazy and will become proficient, probably that's what's happening. We don't know what's happening with machine learning models, but probably that's what is happening with machine learning models. They're both crazy and that's why they they hallucinate and they are extremely good at writing, which is actually in a very interesting way or a sad way for humans, because they're writing so well. We trust them and that's what we call machine endearment, but then they hallucinate. So we haven't really found out actually good ways of using these particular deep learning networks for large language models and DPTs in real sense of the word. Yes, it can write a very good poem for me. It can elaborate something for me. It can write a story, but can I use it in any meaningful manner? And I think this is what Sequoia is talking about. I think this is what also Goldman Sachs is talking about.
Dr.Alok Aggarwal:This is what I write in chapter 11 of the book. Not that they will not be used Again. It goes back to the industrial revolution and its characteristics. Ten years down the road, they will be commonplace. We would have figured out, but right now we don't even have the first, second, third-mile problem, several-mile problem solved for these GPTs. For example, they're trained on Internet data, but a company has its own data, which is paper-based, pdf-based, excel spreadsheets and so on. The first mile is you have to convert all this data so that a large language model can adjust. So those are the ones why I think Sequoia. I think there was also a small comment or a small article from Barclays. Sequoia and Goldman Sachs had been on the forefront.
Dr.Alok Aggarwal:In December 2023, when I wrote the book in chapter 11, I said look, more heist than reality, because McKinsey says it's about 15 trillion dollars will come out of it by 2030 or it will affect 15 trillion dollars. Now the world's GDP will be only 150 trillion, so you're talking about it affecting 10% of the economy of the world. I don't think that's even going to be close to where we are. Having said that, there are many areas which we don't even today. We don't even see where AI is, for example. So many not so many, but many airports, including in the US, have started using face recognition to speed up the immigration process. So in Dubai, if you go, it's your face that it checks and if it's 98% correct, it'll just let you go. Literally, it reduces your time from 10 minutes to 10 seconds. Singapore has done the same. Us government, slightly more cautious, does it all for specific people that it has already screened in the past. Or the global services people I mean travel services people. So we don't even talk about these. Very soon, a lot of these the biometric recognition, whether it's pupils, face recognition hands in many countries not in the US or Europe, but in many other countries will be used as a way to check out of a grocery store. Not in the US, because the US is still worried about privacy, and so is particularly Europe.
Dr.Alok Aggarwal:Similarly, if you look in agriculture, there is a lot going on in agriculture in trying to understand various aspects how much nitrogen is there? How do I implant seeds in the dirt without literally opening up the dirt, because the moment you open up the dirt which has been for the last 3,000 years or maybe 5,000 years that you basically use an animal which has a machine, a small machine in the sense it has a small tool in it which will open the dirt, and then you throw seeds and that always gets carbon dioxide into the air, that always loses moisture into the air and, above all, it loses topsoil to the seas. So a lot is happening in agriculture, and agriculture, I believe, in 25 years, will fundamentally change. So there are areas where things are moving and moving very rapidly.
Dr.Alok Aggarwal:One area which I'm particularly interested in, and particularly because of Scribe, is intelligent document processing. There is so much paper and PDF documents, but these are either unstructured or semi-structured in nature. You need to convert it into an electronic form to be able to automatically use AI, to play with it, to give decisions, support and so on. But unless you can get that completely resolved with very little human intervention, you will not be able to solve this problem. And that I mean out of the 1,000 use cases, there are about 120 use cases on intelligent document processing alone what all can do or should be able to do.
Dr.Alok Aggarwal:Some of them are already in some form or the other, are already being implemented, the simplest ones.
Pankaj Agrawal:Got it. So our audience has a large chunk of our audience are founders who are already building or kind of thinking of building a business right. So what areas within AI right, Do you think a new entrant can contribute to and, in the process, build a large business? Right, I mean, what should they keep in mind while serving incumbents?
Dr.Alok Aggarwal:right, I would suggest they look into three areas very particularly and particularly because generally the entrepreneurs are technologists, they miss out on these areas quite a bit. One is to make sure that you start small with a small problem. But the area effectively. You know that there are many adjacencies, so I'll start with the use case. But there are many use cases which are very close by and I can solve them in the long run. Now that doesn't mean I will solve them on day one. It may take a seven year journey, five year journey. And, by the way, I mean, unless you're an entrepreneur, unless you're planning on selling the company for 100 million or 50 million, most companies take 20 to 25 years really to build. So you should keep in mind as to which particular direction you want to go, whether you're in it for about half of your working life or you're in it for one or one third of your working life or only for five years of it. Both are perfectly fine. I'm not advocating one versus another, but that's something which is very important. Having said that, that they should keep adjacencies in mind, that I'm going to look at this use cases, but use case to make a minimum viable product, but then I'll go and expand it, because those are the use cases that can expand to. Second thing that they should be very careful about is AI is not machine learning and machine learning is not AI. There is no equivalence between AI is a superset of machine learning.
Dr.Alok Aggarwal:As I said, the second AI boom and bust happened because of expert systems. No matter how much we put in, even in LLMs and GPTs, how much we train them 400 million pages, 500 billion words they still require context. They still require and humans do. By the way also, I mean I'm not into accounting Somebody comes to me and says can you figure out what is the total operating income of a bank? Here is the report and suppose that number is not given. I may not know what are the formulas I will use because I don't have the context and obviously AI suffers a lot from that. I mean GPTs and LLMs. A child will suffer a lot, even a human will suffer a lot. So I think it is very important for them to understand the distinction between machine learning and say look, I need to add actually knowledge matter into it, subject matter, expertise. So that is a very, very fundamental thing. My own view is 90 percent of companies startups will fail because they do not include the subject matter expertise into the entire story.
Dr.Alok Aggarwal:The third is not to get caught up with the hype. A lot of people are just taking either open-source LLM models or just open-source models, putting them together and creating a minimum viable product out of it. Great, you can do something, you can create a chatbot out of it, and so on and so forth. If you're a services company, it may all work out, also because you'll move on to the next project. But if you're in a product company, then you don't have any intellectual property associated. So how do you sell?
Dr.Alok Aggarwal:I'll give you an example of intelligent document processing which basically takes the data, converts I mean which is scanned data or PDF data converts it into electronic data. But unlike humans, which have two eyes, the machine converts it into literally one dimension. It loses the context of tables, loses the context of graphs, charts, etc. They have the second problem, which is to recreate tables with pretty much 100 percent accuracy, graphs and charts. Then suppose I come and tell you that look, it's 90 percent correct. You're my client. Your immediate reaction is okay, that's pretty good, it's 90% correct. You're my client, your immediate reaction is okay, that's pretty good, it's 90% correct. But this is a 300-page document you just converted. Will I have to review it from left to right for all the 300 pages, because I don't know where the 10% is wrong? So the question is if I can reconcile, for example, various suppose there is a table in it which is income state, I can reconcile and I can show you that look or not me. But the software says look in this statement, a plus B plus C equals D. Then you don't have to review it. So I can very, very clearly point out which are the issues, where are the issues, and so on and so forth. This is just one example that you add, because everyone I mean there are 57 or 58 companies that we know of. There may be more companies in intelligent document processing, but none of them.
Dr.Alok Aggarwal:Everyone will say oh, you will need a human in the loop. Not clear that you will need a human in the loop. In humans, yes, we have a maker and a checker, but it's not that we use checker all the time because checker is expensive. It's our two eyes which solve the problem. With my right eye closed, I look at you and I say it's punctured. With my left eye closed, I look at you and say it's punctured.
Dr.Alok Aggarwal:The error of my being wrong is squared, because if my right eye was 90% correct, error was 10%. My left eye was 90% correct, error was 10%, and if both eyes were working independent of each other, the error becomes squared. That is only 1% wrong 10%. Wrong there 10%. So what I'm trying to say is there are interesting areas that people can go into. I'm not saying everyone will, but I'm saying that in product business you have to have intellectual property that you can defend and unfortunately I do not see that right now. This is one of the things where people are rushing. It's more than a gold rush I'll call it a platinum or a diamond rush that people are not rushing to figure out what the issues are and how to solve them. Rajesh.
Pankaj Agrawal:KASTURIRANGANANI, could data be that intellectual property? So one common theme, or rather the playbook and I wrote about it on LinkedIn recently that a playbook is, as you said identify a painful and a use case, build a superior product to solve that. That becomes your sort of foot in the door product generate demand, bring along customers. Doing that In the process, hopefully you are generating some high-quality and proprietary data which you can use to kind of keep on making the product better. Do you think the data itself could be that IP?
Dr.Alok Aggarwal:I don't think data will be the IP. I think use of the right data is part of the design knowledge and that's where subject matter expertise comes in. So what data will I use? Because there is a lot of data around, but a lot of data is also noisy and you can't use it.
Pankaj Agrawal:Quality of data.
Dr.Alok Aggarwal:So that goes back to subject matter expertise. What data do I use? How do I get to creating a product which is superior? And when I say superior, it has intellectual property which is hard for people to overcome.
Dr.Alok Aggarwal:The sad part about AI is or, in general, software is yeah, you can write patents, but they're worth not even the paper they're written on. You give me a patent in software, in AI, because it's math. Eventually I can go around it fairly easy. It's not like building a company, and I know you folks are in the venture capital business business and I know many VCs who get all excited about patents. Even though we could probably in our company write about 30 to 40 patents, we have not written a single one because we know the moment we write a patent, we have essentially given our intellectual property away. Somebody will just go around it. Yeah, it's like reconciliation of intelligent document processing. We have not written a single patent on how do we reconcile? How does the system learn formulas which are there in a PDF document automatically, or how does the formula get learned in a football program?
Pankaj Agrawal:I think rightly so. Another theme which I agree with that you pointed out that subject matter expertise will be relevant. So if you see, now Lama came up with 405 billion parameters and it is kind of giving a chase to GPT-4.0. And a lot of experts are kind of saying that GPT 4.0 is like a Ferrari for grocery shopping. Right, it's just too over the top for most of the things. Right Versus open source model that you can access, you can utilize, you can train on your specific use case, make it like you know, work really really well and accurately for that domain. So that will be the future and that's where subject matter expertise becomes more and more kind of relevant. I personally feel that models themselves are probably the fastest depreciating asset in history, probably.
Dr.Alok Aggarwal:In some sense they're not even depreciating because they become open source very quickly. Like Lama is open source, I mean Mixtral is open source. I can choose a bunch of open source models and then I can put them together. Now, what is intellectual property is? Lama has, let's say, in a particular metric, 90% accuracy, mixtral has 88%, and all of these are open source. Can I now put them together in a meaningful way so that my accuracy becomes 92 percent in the whole process? That becomes my intellectual property. Now, if I put subject matter expertise on top of it, then I can get to 95 96 percent and then if I can reconcile everything and show you that there is no hallucination, because I'm showing you where the where the particular thing got its answer from, then you are a happy person as a client, right? I mean then that you say look, you have really saved me time, money, cost etc. And above all, human labor.
Dr.Alok Aggarwal:So I think these things will take time. They always. And that's why chapter one of the book is so important, because it sets up a framework that, look, let's not get carried away. I mean it's perfectly fine if we get carried away. Also, because I do say that hype, boom bus cycles are actually good for society. I mean, we saw that with driverless cars recently. About $100 billion was spent in there. Yeah, some amount of it really went into research and eventually we will solve the driverless car problem. It may be eventually, maybe 10 to 15 years from now. So I'm not even saying that cycles are bad cycles For the human society. They are actually probably good things to have. But investors and inventors of course, they lose their shirt, so they feel bad. Yeah, that's it.
Pankaj Agrawal:This has been super interesting.
Pankaj Agrawal:I have one last question that I've been kind of trying to unpack for the last couple of months as I go deeper into AI.
Pankaj Agrawal:I think AI, with its power to, you know, either as a co-pilot or you know, kind of make the overall system more efficient, has the ability to kind of add meaningful value in services businesses right, and there is an amazing concept that, rather than software as a service, there could be certain areas where service as a software has the right characteristics of a venture scale business producing venture scale business in terms of capital, efficient growth, velocity of growth, and so on and so forth. Given that you have experience in both now with E-valueServe with respect to KPO, and now with Scribe, you are buildingalueServe, you know with respect to KPO, and now with Scry you are kind of building products for these customers, large customers what is your view? I mean, if you were to, let's say, think of starting E-ValueServe in today's world, would you do anything differently? You mentioned that it took you twice the time you projected to get to the scale right, or where do you think? How should an investor think about the services space generally and how can it be powered by that?
Dr.Alok Aggarwal:Actually, after writing the book, I decided this year to convert two things To convert the book into 36 lectures of 45 minutes each, because not everyone reads books these days, very few people read books. And by no means the book is small. It's actually 270 pages of fairly dense material. It's not intense because it doesn't have math or computer science, but it is dense. You have to think about it, read it once more before it begins to seep into your into the brain, into your thought process. So, uh, so that's one. And the second is to write at least eight articles on India and AI. Because everyone talks about China and US. People are forgetting that India has the largest number of software professions right now in the world 30 million developers out of 30 million.
Dr.Alok Aggarwal:And that is 5.5 million software developers actually working on a daily basis in all these companies. And we're not talking about HTML and CSS developers, we're talking about software engineering developers. I mean, the world has about 25 million and India has about 5.4. So people forget about that and that is going to be a very, very important aspect about India's growth in AI which people don't realize. Having said that, so my first chapter was first article was the unsung heroes of AI how data annotation will grow enormously, meteorically in India. The second one was, of course, loneliness in India will rise because of AI. The third one is that the services industry in AI will be transformed radically by AI in the next 10 years.
Dr.Alok Aggarwal:Because you talked about EvaluServe. When I think about EvaluServe all the time, because I'm still invested in it fairly heavily, even though I'm on the board, and if I were to redo it, I would just completely redo it with software first and humans later. And about 50% of the work that EvaluServe does and, to me, about 70% of the work that TCS and others do, especially in the BPO domain, like GenPan, like eXcel service, 70% of the work is within the next three to five years can be reduced by a factor of two which is a lot right, Because suddenly clients begin to realize that you've reduced it by a factor of two, they'll say why are you charging me X when you should be charging me X over two?
Dr.Alok Aggarwal:And then you say, okay, there is a software cost. They'll still pay you maybe 0.6, 0.7 times X, so everyone gains in the whole.
Pankaj Agrawal:And this it can also expand the overall demand. Right, the Jevons paradox will kick in. You reduce the cost of technology per unit consumption, the demand will grow up.
Dr.Alok Aggarwal:Absolutely, because second one is it will go pay by the drink, not by the FTE, so to speak. So you will go into, as you rightly said, small, medium-sized businesses who cannot afford any Excel or any value serve today will be able to afford the next generation of companies.
Dr.Alok Aggarwal:So I think, this is going to be a very, very fundamental change. The only good news and this paper should come out in September the only good news I see is that it is not again not going to happen in the next three years. It may take seven to 10 years for this change to occur, but this change is imminent. This change is. I mean I hope TCS, infosys etc are waking up to this change. It's not that these people will go away, but I mean you will still have 50 percent. You'll have people who will go upstream and so on. India may actually gain a lot. It is an opportunity for India to really play very strongly in the industrial revolution, because the foundation is set, because India has so many software engineers, because India can produce enormous number of data annotators who can check for ground truth, who can basically do data annotation, who can tell you that look, this part is this portion, because in supervised learning we need data annotation.
Pankaj Agrawal:Fantastic. Thanks a lot, Really really appreciated doing this conversation. Of course, we'll recommend the audience to check out your book. We'll link it in the show notes and thanks for taking the time to chat with us.
Dr.Alok Aggarwal:Thank you, Pankaj. Thank you so much. Thanks, Jerome, for making it work all of it. Thanks, I'll write to you separately.
Pankaj Agrawal:Sir, I'll take your details from Jerome. I would love for you to talk to Empirical and at some point in time, of course we would want him to spend some time in the US, so of course we would love for you guys to. I mean, he's from your alma mater, so that's alma mater, building something in AI. So I think two strong reasons for you to take the time out and have a chat, and I think it will be helpful for him as well.
Pankaj Agrawal:Yeah, sure, absolutely All right, have a great day, sir. Thank you. Thank you Bye, bye.
Prime Venture Partners:Dear listeners, thank you for listening to this episode of the podcast. Subscribe now on your favorite podcast app for free and you'll be the first one to know when new episodes are available. Thank you, we would be really grateful if you leave us a review on Apple Podcast. To read the full transcript, find the link in the show notes.