Kailash Nadh, CTO Zerodha, on the paradox of A.I. | CRED curious with Kunal Shah

there was a time where technology could take Blue Collar jobs this is the first ever attack on White Collar jobs and it seems like it's a fairly deadly one uh right now people are like okay I'm just more productive I'm doing more stuff I'm editing some stuff here and there let's talk about GPT 4 plus plus or and that's not the only llm that's going to be out there competing maybe there are some of them you don't know that it probably exists in some ways what happens like just imagine a few scenarios let's talk about jobs and jobs are cred it's a very touchy topic it's a very politically sensitive topic everywhere put a disclaimer in the video and start [Music] thank you for everybody who's here and everybody is going to watch uh you can pause and stop if if you not don't like dark stuff it's gonna get dark uh anyways I'm gonna start uh you're gonna chat about what's up with the world and the impact of AI we'll talk about climate change in conjecture conjunction of this whole thing and we'll also make sure that uh we try to keep it evidence based database and not feelings based and and will not challenge each other all that stuff will not do okay uh let's start with this thing about what is AI and when did it start and why is this different this time I think despite all the technologies that we have people still cannot agree on like a single definition for AI there are many many different definitions with the simplest definition is they attempt to recreate human-like intelligence in machines when did it start I think if you really look at it it started many a millennia ago it's it's not a new phenomenon it's not two or three or 10 or 50 years old because this is one particular technological feat or attempt that has deep roots in philosophy and some of the debates right now are extremely philosophical as they should be uh just I think it you can trace it back to this uh to the debates of Mind versus Body Mind Over Matter mind and body being the same or diplom being pure material matter and that's thousands of years ago and the whole matter versus mind debate raised in Ancient India in different Indian philosophical tradition Traditions it was uh like a big Hot Topic in ancient Greece I think atomicism the idea that everything everything arises from my new particles called atoms that's I think yep you can really trace the roots of artificial intelligence all the way back to this debate and coming to more practical relatable terms this whole idea of automatons humans building lifelike dolls and with cogwheel mechanisms creating lifelike Behavior there were Bots that wrote Set script you know once wound up in I think Medieval Europe in 19th century etc etc so these were all I see all of these as uh all of those things as attempts towards trying to imbibe human-like intelligence in machines so what we have right now the modern uh AI items again that is also that predates our idea of computing and the modern computer science in general like we would like we were saying earlier I think the first attempt to mathematically model a human neuron to codify it happened almost 100 no more than 100 years ago uh French mathematician and then I think 30 40 years later the first artificial neurons were created I remember when I was in university We we played around played around with some of these Concepts I think Michael lope its neurons early 40s and this was before we had the modern computer architecture and these were physical devices that behave like neurons you know electricity goes in pulses come out after a threshold so that is I think that is the pivotal moment where the modern kind of AI starts and that is 1940s before computers before modern computers so then there were plenty of breakthroughs in the 50s more neurons uh more kinds of neurons uh were created mechanical neurons then you had once we had computation software neurons made it easier I think perceptron was one one of the first serious attempts that really kick-started the modern AI research that was in the 50s and then plenty of things have happened and we've arrived at let's say Transformers deep learning etc etc in the 80s there were pop field networks which I think which were the starting point of modern neural Nets so plenty of Milestones but the point is this is not new the breakthroughs are new but the things that have led to these breakthroughs cannot be looked at in isolation they are they're a result of 70 60 70 80 years of actual modern mechanical electromechanical computational research in the philosophy that has resulted backed all of this up is literally thousands of years sorry that's like a quick I know I know I'm gonna pause somebody's playing some English song needs to stop is it you is it this TV it is dead there's a faint background I know I think the song also like it's a it's a Halo somebody singing in that song oh it's my phone music do you listen to Beyonce is amazing who doesn't like Beyonce it was Beyonce I I was like familiar with a few strangest thing right I've gone on like three or four floors here the Terrace here and there was this faint background of music like you must have some mechanism here you are the mechanism s can we edit Out Beyonce from the real it's just for some sanity over here okay I'm gonna go straight into um uh what is different this time uh why is this time exponential uh we will not talk about why humans don't get experientiality that's understood we've talked about it many times that how we forget about we don't understand compound interest forget about exponentiality so let's just ignore that yeah our brain cannot process it can you talk about why is this time exponential and and can you give some examples so when people say there have been breakthroughs to define a breakthrough you need a baseline if somebody does something out of the blue incomplete isolation you can't really Define it as a breakthrough so why it's different this time you know the quick primer that we did you really have to consider all of that many a millionaire worth of thought you know a century worth of research all kinds of artificial intelligence Technologies and breakthroughs and researchers have happened in the last century and how what's happened now over the last couple of years or maybe the last six months how it's different so it's evident once you if you really look at what has been achieved in the last 60 years with computational artificial intelligence and what's happened in the last six months it's evident from its Its Behavior so the Holy Grail was computers understanding in quotes human language and that that's one of the biggest Pursuits was natural national language passing entire computational field of study but now there are these things called llms where you don't have to put in rules of natural language passing these things can pass and respond uh even reason or there's an illusion of reason at least in all pretty much all known human languages pretty much all known computer languages so it's just kind of absorbed the entire field of NLP you don't now really have to do natural language passing algorithms there's this one thing that does all of that then the other thing was of course illusion of reason right when you ask it something you should get something meaningful back that's happening and forget that you can have a Chain of Thought conversation back and forth throughout so these are all entire fields of artificial intelligence over the last 60 years each one of these you know kind of behavior and and conversing is just one tiny piece there there are these things called I mean it's called emergent phenomena that's the simple term there are many different kinds of weird behaviors emerging from the same model you know like a large language model yes language is the way to interact and communicate with it but the way it there are examples of these some of these models attempting to reason getting into logical arguments you know solving logical quandaries so there's this one thing one class of Technology called the llm that is showing n number of behaviors that has kind of subsumed entire fields of AI research all into you know this is one thing and that is why it's different talk about impact on human jobs for a second we'll go to humans later on uh uh there was a time where technology could take blue collar this is the first ever attack on White Collar jobs and it seems like is a fairly deadly one uh right now people are like okay I'm just more productive I'm doing more stuff I'm editing some stuff here and there let's talk about GPT 4 plus plus or and that's not the only element that's going to be out there competing maybe there are some of them you don't know that it probably exists in some ways what happens like just imagine a few scenarios let's talk about jobs and jobs or credit so it's a very touchy topic it's a very politically sensitive topic everywhere put a disclaimer in the video and start it's okay to talk about it but it's a very touchy topic and if we say that when you say these this class of Technology will cause widespread job losses uh that's a blanket statement and it may be true it may not be true but I think we have to it's better to look at it step by step so that people who do not see the risk also see the risk I think it should be it should be serious it should be a series of Sherlock in logical deductions uh we can we can work with simple examples so the entire point of the Crux of the matter of an llm is language and it can consume it has already consumed large amounts of text and knowledge uh you can feed it large amounts of text and you can ask it pertinent questions in your language Japanese Hindi English doesn't matter it's one technology and then it can quickly uh sift through the semantics the meaning of the let's say 50 pages for the text that you put in and give your pointed answers now that is a job typically a researcher would do you'd give someone a 50 page legal document and say you know what I want to know these what are the pertinent points they'll spend after day one day two days whatever they'll do their research and come back with a summary now there's this thing where you can feed it huge amounts of text and suddenly get summaries can't guarantee that it's right you also can't guarantee that humans are right but that is just you know that's one entire class of jobs it cuts across Industries at credit there'll be people who consume information to produce intelligence and zero there are people who consume information textual information to produce intelligence every company every industry has this role and so logically if there's a system that can do all of this within seconds and it's let's say even 95 percent great logically uh how does that affect the human jobs that were doing this and this class of jobs alone is probably millions of you know people across the world so simple logical deduction and just that's just one this is already possible yeah let's talk about possible what will be possible we will try to hint at AGI and or the past to AGI what starts happening so what is all I think we have to really look at the entire idea of end order effects this is a simple first order effective text in intelligence out and so imagine I'm just you know I'm not saying this is what's going to happen it's just a thought experiment imagine uh 10 people researchers who are doing this suddenly you have this one model an AI model that now produces intelligence where does that intelligence typically go in a let's say for-profit organization that intelligence will be used to make certain decisions you know conditions and that decision if you think of it from a first principles perspective let's say you've got five bullet point uh key points from a 50 page document The Next Step would be to read those five bullet points and take an action you know send an email or click a button on a CRM whatever so what would be the next logical step you know have bring another system in that takes this intelligence and executes a small little action in the beginning it would be low risk actions but if that starts working what would be the next logical advancement no slightly higher risk actions to the rather than trying to rather than trying to Envision all kinds of things this can do I think it's which is impossible to predict and thought or if it's easier to do first principles dissection from just this one thing of text summarization it has the potential to move on to decision making and once you get to decision making that's a whole different ball game yeah from low risk decisions in an organization to I mean if that starts working you know the risk is slowly upped until you systems automated black boxes start making decision so if there are systems that can do decision making up to you know there could always be a human that's finally says yes or no then you can imagine the impact on every single human organization where I mean everybody runs on decision making it's interesting so I'll talk about a story uh which is like a nice precursor to what can be the type of job so one of the most highly paid job in the US is of a radiologist to reads x-rays MRIs all of that and and once upon a time when one of my earlier authors I tried to build a tele Radiology startup which was basically Indian Radiologists who cost marginally uh like 100 the potential cost of a U.S audiologist could do an assistance program and give that for the U.S audiologist to kind of approve yeah that okay this is a good enough detection and obviously kept the rare cases for the main radiologist to see because 90 of excess mri's turn out to have nothing in it now uh what was interesting is there are three things that were shocking the the U.S radiologist made comment that the Indian neurologists are actually quite good like what is going on and the the simple thing that they came up with the Indian Radiologists just saw so many MRIs in x-rays because our volume of country is huge and our doctors per capita is slow they are just like really good at their own AI so they were actually quite good second is the cost was dramatically lowered the uh the third thing that was interesting is that they were actually doing much better on the rarer cases and the misses so some of the reverse thing happened they said why don't you audit my last 500 readings and see if I have missed anything and the identiologists actually did a better job of detecting the passing now obviously the startup did not go where they were regulatory challenges I was not sure about the market size all of that is separate topic but now with AI I don't know Indian or U.S Radiologists will be able to stand a chance maybe in five years from now by the way if you don't know what's going to happen in five years we've spoken about uh it's impossible to predict what's going to happen in five years but these are not even the category of jobs just stuck on texts yeah processing and all of that this is years of compounding knowledge of becoming a radiologist and going through so many things and being able to diagnose Well yeah if diagnosis goes away Solutions can be playbooks because we basically that's what medical science is in some ways obviously the rare cases will still go but there is evidence now that rarer cases are being solved by AI at a much higher accuracy so we are entering some scary times now on that positive note uh what's the 10-year View what should we be doing I I honestly can I can I ask a personal question yeah you've chosen to not have a kid yeah uh and and it was not an emotional decision it was a thoughtful decision I'm assuming uh and and you can change that decision if you want but what was the background of this and and is it correlated to what's going to happen in 10 years so that was a it's a very difficult decision as one can imagine especially uh in the Indian context took a very long time to arrive at that decision and of course with the full participation and consent of my partner but it took probably I know eight almost a decade to arrive at that decision and the seats really were planted I think in the late 2000s in me and it's really to do with climate change so this whole scare that the world is only going to get worse so it's not ideal to bring a child it's not right to bring a child to the world is probably again 5000 years old because the world was way more violent and chaotic and disease ridden you know you're many many the most peaceful time ever exactly so in absolute terms you are living in the most peaceful time ever highest human life expectancy everything's gotten better uh so that's the dot but if you really pay attention to what's been happening with climate change you know the amount of destruction uh that's his first order in the order effects we don't even know over the last maybe 30 40 years when it really accelerated I think we are moving to a world let's sorry it's not my thought there is mountains and mountains of evidence out there that we're moving towards a world where many of these unpredictable climatic events could just be catastrophes which will displace a really large number of people and who will get this place as always the you know poorest the most marginalized people who subsist on small little wages every uh everyday people who have limited access to water so we we just have to I mean we've all been reading about heat waves right the super erratic heat waves uh the ice caps melting hottest here in the history uh everywhere from you know rural villages in Kerala to parts of North India to the poles so if you just if you really look at the evidence we're going to be in a space where there'll be increasingly chaotic and climate is a feedback loop system like what was happening with that you know Mike and speaker speaker earlier right it's a feedback loop it accelerates so we're going we're going to enter an era we've already entered an era where these unpredictable events you know erratic Monsoon erratic Summers erratic monsoons imagine the nth order effects it affects huge amounts of Flora and finite effects are food security all kinds of things and that's just real or or water for them yeah yeah and there are you know questions about water conflicts there are you know yeah yeah signals are emerging there are islands trying to island nations trying to move themselves it's these are all simple examples but the nth order effects of such a complex climatic system which has emerged over millions billions of years which we managed to mess up with our Industrial Revolution and technological progress in the last 100 years is going to be really nasty so it's it was entirely based on evidence and it seems it just seemed logical that it's illogical to bring yet another child to this world where there'll be a large number of marginalized children suffering and we're all privileged everybody sitting here is privileged and we'll figure out we'll buy ACS we'll do whatever we'll do something yeah yeah the point one person will always has will always figure something out but it just seemed wrong so and that's just climate change you know there's this whole big litigation happening in the US I think forever chemicals I don't want to name this large corporation but they've emitted lacks of different kinds of forever so-called forever chemicals over the last many decades all over the place which permanently cause damage in the forever because they don't undergo reactions and dissipate and they're being sued but this whole thing about you know suing what does that even mean in the grand scheme of Earth's history we are here in a fleeting moment and a bunch of humans are bickering so that's that's been my reasoning uh on that particular topic let's just move back to thinking about AI right now uh you talk about the point one percent privilege right and and we've always been talking about world has disparity of wealth uh but it always existed humans never had this concept of it's a it's a almost like physics it's always going to be true uh so while we increase the base poverty limits and we got more people to kind of be okay uh the disparities only increase because the total amount of wealth in the world had dramatically and exponentially grown since Industrial Revolution right yeah I feel it's going to be the making that you remember the chart how the GDP of the world has gone like this yeah I think it's gonna go two levels we cannot imagine now right and at a much faster acceleration rate which means the disparity of what we have felt so far is going to look like a joke now because the richest set of 500 human beings are going to be at a different order of magnitude of wealth and power right now the question is that if this is going to happen and we always had this conflict of disparity and all of that what happens to the concept of Nations what happens to the concept of conflicts uh uh how do poor Nations survive how do rich Nations benefit how do you see this evolve and the answer to that is going to be extremely dark but again we can play that Sherlock in Game of logical reasoning and deduction uh when we speak of Nations and borders and GDP these are all brand new Concepts just a few decades and before that there was no codification of GDP yes and the geographical Maps looked completely absolutely different yeah so I think it's even meaningless to talk about these Concepts that have lasted barely a century yes and if you look back and beat back in history maybe even 3000 years ago if you pick let's say Carthage That civilization that I don't know for some reason the entire world has forgotten but Carthage lasted I don't know 700 800 years and it was one of the most powerful civilizations uh on the planet and within that Civilization there were periods many eons of dictatorships aristocracy republic democracy same thing in ancient Greece same thing in Ancient India all kinds of Kingdoms have gone through all kinds of Cycles yeah so we when we speak of today's world geography Etc we are ignoring you know three to five thousand years of other recorded history and these things have been having been played out countless times you know in ancient Greece there was a period where you had 50 years of democracy 50 years of Republic suddenly became a dictatorship and suddenly something else if we Peak back it's just a distant blur yeah but our present right now it's only what just a few decades right so how can you even I mean it's meaningless it's meaningless to project these what could be temporary things GDP lasted the idea of GDP was formulated in what the 30s barely essentially into the future they might the concept might disappear 10 years from now that's how humanity is always money concept of country we could keep redefining I mean who knows right so it's just so it's very delusional of us to root our entire idea of future in the last 60 years we really have to read history well that's a very important topic most people have not really read the history except what came in their academics which have been obviously uh influenced through what the country wanted you to know uh if you really go to history and see patterns of it so you talk about civilizations um there's an interesting study and books and YouTube and documentary for how civilizations end right oh yeah and and and I've always uh like so I between two companies that come up with Delta IV as a concept and it was actually born out of physics and biology on how a species wipes off another species and basically the key Insight is the new version is significantly more efficient at managing energy which means keeping their entry people's local entropy low and overall increased entropy for example air conditions is keeping our local entropy low right we are not in struggling and the increasing the external entropy by passing on this heat and dissipating it further now if you take this concept further countries are nothing but these complexities again species are these complexes so we call it the entropic complexities which is just nothing but storage of energy is going from one structure to another structure so the study that which is also released point to one thing that civilizations are nothing but a store of energy which displays into the next door of energy basis their ability to be more efficient at doing things right and and therefore we are going through this interesting transitionary thing where the the ability to create Delta 4 is now a lot more easier than what it took the amount of energy it required so therefore a lot of people make this joke about can you make actually I'm going to ask you that question can you make an llm for I'm gonna give you a hundred billion dollars can you do that so uh that you know that reminds me of this very famous quote on Carl Sagan so uh in his I think one one of the episodes of Cosmos he said to create an apple pie to make an apple pie you must first create the universe from scratch and that is a legendary absolutely mind-blowing quote what that really means is that nothing really exists in isolation so like we discussed earlier today's AI the Transformer breakthrough after 2017 of the llm GPT breakthroughs of the last two years Etc couldn't have spontaneously emerged you've had to have the entire civilizational progress Etc that have led to that moment including actual work research Etc so if you ask me can I uh I'm firstly I'm not an expert in llms or gbts but if you ask me there's this x amount of money can you create this really complex piece of technology out of nothing the answer is no so how can one create something like that out of thin app you can't create it out of thinner open AI used the research that came out of Google brain project to build this now if Google hadn't published that paper we most likely wouldn't have had these breakthroughs we wouldn't be chatting about this so that the thought that you can create something extraordinarily complex out of isolation like one person or 10 people is largely delusional I would say if there are people out there who think that I can you know put X amounts of money here you know let an llm emerge it it won't so it worked out for open AI because they had the right environment where like we discussed again there were decades worth of research people culture DNA access to material access to technology access to money it's it's a series of events that have led to the why was opening AI founded even that is a you know yeah it's a it's an event in the Series so sometimes it's possible to create things out of no out of nothing like wartime Humanity changes when there's less a threat of War people suspend their entire lives come together build extraordinary things but that is a exception that's an exceptionally rare event that's done no that's not the norm so in peace time if you say I'll give you X build y the likelihood is really really really really low uh-huh I'm gonna switch back to a personal situation now 10 years we have should we be doing all this like job building companies or like be chilling because I don't know what's going to happen in 10 years what's your recommendation on this I genuinely think about this The Show Must Go On Right and that's how it's like the Titanic guy who like playing a great historical fact really I mean what can I don't know I mean if you think about it forget AI the threat of climatic Doom planetary catastrophe hasn't caused us to pause and for people to collectively try and save the planet I mean air is cool right we have we're still thinking it's cool right now yeah I mean if there are cool products coming out right so if the threat of planetary collapse has in United humans we are starting large-scale Wars right now as we speak yeah I think I think what AI can be used for when it comes to military is a different beasts all together uh let's switch back to uh India um let's say we were to really give recommendations what should India do what Indians do what would you say what to do I I find it extremely difficult to answer questions where we sum up India you know such a complex civilization 1.4 billion people into anything you say you're probably trying to project yourself as a as a representative of the entire concept Nation idea of India it's very difficult I I don't know what is right or wrong India is not one homogeneous entity it's one of the most complex countries in the world how could you possibly have this one recommendation that takes into account all these complexities in such a non-homogeneous society I I have no idea but we can we can try and experiment we should Tinker that's what are you going to do for 10 years what's your plan there is preair and there is post AI what are you doing what has changed in your life and your choices in your decisions in your long-range planning I don't have long term plans that not in a nihilistic sense where I don't care and I'll do whatever that's not how it works but I find it impossible to peek into let's say two years from now let alone 10 years from now I mean forget I can't even know I don't even know what I'll be doing next week how could you project what a billion people will be doing 10 years from now I don't do that so I try to do whatever is in my control there are certain things that are in my control what I choose to say do the projects I choose to execute you know so my choices are largely limited to that what I can do so I uh I'm involved in way too many projects so I pick projects that I think are that that you know produce societal good it could be a small technology thing could be some of the work that we're doing via you know our foundations could be any of those so I'll continue doing that and it doesn't bother you that you can barely make an impact to the yes it bothers me that's that's the exact existential crisis right once you realize that or maybe once you accept that no matter what you do perhaps nothing will change then you know you go into that trap of defeatism and uh but that's a very dangerous trap but that is what it is to be human right despite knowing that it might not make a material difference we'll still try and you don't know what is going to compound and you know come come out of all of this so like I said we can only really worry about what's in our control so I try to only worry about my actions and the outcomes are entirely random I mean I'm I have a very I have an absurdist worldview uh a lot of things that have happened in my life and like let's say last decade have been complete coincidences for example is a complete coincidence we happen to be on opposite sides of the road all of that oh is that true is that what happened yeah I don't know what the story what happened how did you mean so it's a chain if you trace that chain of events back we'll have to we'll go back to the single cellular organism but that is why long-term plans I think I'm absurd but I'll give you an example so I was in the university for six years then I probably moved to Bangalore because Bangalore had better internet this was 2012. and I was here working on my projects you know technology stuff hobby projects and whatnot and an old school senior from uh sorry an old school senior reached out messaged me saying oh I remember that you used to fiddle with computers in school and we spoke for the first time after a decade you still do I said I still do he's like I have this idea can we catch up and he happened to be in South Bangalore and I happened to be in South Bangalore and I happen to be in South Bangalore because when I wanted to move to Bangalore I had no concept of the geography here and my cousin was studying in South Bangalore so they're like oh he's there so I'll also moved there so even that is a coincidence you know over a phone call so we met up we caught up and uh this is Abid from sensible Sensible is another Financial platform so then we he wanted he had this idea of building this financial app and I had no clue of Finance back then possibly even today not something I'm interested in and I got interested because it was a great engineering challenge 2012 there were you know Goods there was good stuff coming out high quality consumer software technology and I looked at this whole Finance landscape and there was nothing happening and to a programmer to a software developer that's like a that's a very exciting prospect like oh you know Finance I'm really excited to do new things let's do this so we started working on this then we had to look for financial organizations to partner with and we were trying to speak to all the Marquee names nobody really bothers and South Bangalore and we used to hang out at his place and nitin's first office happened to be like 200 meters away from this place so one day you know it just whatever happened he walked in said hi and that's how it started so interesting but a string of coincidences yeah a lot of people say that to build good products you'll be the user of product do you do a lot of trading buying stocks mutual funds I've never traded in my life I've never I'm not a Trader or an investor but that's that's why you have teams that's why you have collaboration we all I was just curious I was just because I I I expected this answer so I was like curious but were you great curious the story of curious I I was thinking about how did I get to this curiosity thing and I was like thinking about a coincidence and I'll tell you my coincidence story when I was in first standard my teacher gifted me a book and because the character in that book looked like me and the book was about a curious kid so I just assigned this personality to me because I had to be that person in that book yeah complete coincidence and I was just thinking about like how did this whole thing of being curious happen yeah and and I was just like I could be just like not that my family has some crazy academic genius in thing or not that I have like have people around me who were like brilliant what happened and I was just thinking about this identity that I assigned because of a random gift that my teacher gave in my first time that that's how humans have always lived their life that's how big and small events in history have always happened that one phone call that one conversation that one word that you owe here completely changes timelines so and the butterfly effect right that one tiny thing that happened 30 years ago that left an impression on you can completely change your trajectory and you end up going and affecting a number of people that changes the course of history in so imagine this happening with eight billion humans where there are coincidences shaping so in such a world how could you possibly make a 50-year plan or 100 year plan it's great it's a great attempt but I think it's a nice thought exercise okay on part exercise let's do a thought exercise a large chunk of our market cap is software services companies uh We're Not Gonna discuss what happened to the market cap we're going to talk about uh they are currently providing some Services all of us have either relatives family members close siblings all of them are connected to the industry and we are somehow part of the industry even though we may not be working on these companies yeah how are they going to get affected in your view and this is a thought experiment so I think there's a we can have a nuanced take here there are many different classes of programming or software development OR tech jobs not all programming jobs are the same there's a whole gradient and IT services the way we understand it in again in the Indian context it's a certain class of jobs actually before that I I spoke to ended up speaking to someone today who said that they spoke to someone else today and this organization had an opening for six developers and this is a product tech company I think and they this gpt4 stuff whatever co-pilot whatever they're using apparently turned out to be so good that they've decided to close those goals they no longer hiring for those six you know positions and there's just so this has started happening right around us yeah in small Pockets so sorry coming back to IT services ironically I think maybe it services the way we are looking at it may be shielded from this for a little while because again first principles thinking uh when you Outsource large amounts of your technology it's typically old ancient Legacy technology uh you know 40 year old technology and stuff yeah yeah yeah yeah yeah so these are so complex these systems are so complex so old so badly maintained over let's say 20 30 40 years that people are afraid to touch it anything that you change might just you know destroy 40 years worth of Legacy so these these are not good code bases where you can just add developers to you know expand extend you're afraid to touch it and it's business continuity if you have a 40 year old system if you're a large Bank a 40 year old system breaks that's that's done right done for so maybe this is just because of this incidental reason that it's not easy to integrate new things into these old served technologies that we service the technology Services industry Services it's not easy to change Parts it's not easy to plug and plug out maybe for that reason this may be shielded for a longer period I'll challenge you on this uh the host companies yeah that you're building this for they may not be able to survive some Jungle which is not electricity so I'm safe from covert it seems like that right yeah because I'm in this cool tribal place called this fourth Pro right but the one I serve is not disconnected they are not in some private island yeah they're competing in big markets yeah and unless they become more efficient they're going to die right uh and therefore my host companies will want to do something more time to change things so maybe that changes things it's true I mean there are so many variables but there's this whole class of regulated organizations that says Banks Etc or oil companies who will still remain relevant because they're gonna fuel the air does that but I think such disruption will probably not be allowed in heavily regulated Industries so when we speak of uh large-scale organizations outside India availing tech services from India many of them are large Banks Etc so Regulators will not suddenly allow sudden disruption I mean you can't just go disruptive bank with AI or whatever that's not how it works right so there are so many variables but assuming that some of these large organizations are protected by regulations for a while there is some protection yeah either you are an ancient technology or you're protected by some regulator or by a country that seems to be like some safe spaces in some ways right yes for short periods of time because when on the one hand you you have this Legacy system that you can't touch so that you know code generation AI is not coming to you know replace and do whatever replace people who've been servicing it but on the other hand uh there'll be new things that emerge that are far more Cutting Edge and sophisticated so for how long can this entrenchment protect it's just a matter of time but I think you'll be protected for a little it's just a guess okay a little while more the services from all that you've seen how far are we from let's assume it's going to happen but like how much chance would you give it to a marble quality movie being made completely with AI with no humans involved difficult but I don't think it's science fiction at this point because literally every week There's a new model that is coming out that is upping the ante a little bit by the week sometimes every two days there is some Cutting Edge breakthrough happening right so just this uh just today there was a demo I read this on Hacker News I saw it on Hacker News where somebody's figured out things stable diffusion to generate videos and it's just some Hobbies somewhere I think so I think it's not a lot of time it's not a lot of time you could like have four Marvel movies a day like that's a good future right so that that you know that opens up an entire Pandora's box of other philosophical quandaries if you have a button that generates the best possible movie that you like every time you press it I mean then you know we are satiated right you'll you'll not you'll stop caring because if everything is always perfect at and it's everything is always available at the click of a button why would you watch I disagree I'll tell you why I would say it's not going to be the case because uh so I read an interesting study it says that why can kids one watch the same cartoon movie ten times and laugh every time right because the human brain takes some time to detect patterns and predict so for them every time it's exciting experience but we need Christopher Nolan level stuff to be even think about watching it again because the next time will be like yeah I can't laugh at this I can't watch this again I know what happens next all of that now unpredictability through AI in terms of scenarios like you can make a Black Mirror episode almost every 10 minutes if you think about the variability it can think about with the second order effects third and hundred effects can be such interesting possibilities right for example you know one of the things that I've done many times in interviews at credit I asked somewhat morbid questions but just to test second order thinking is like what if everybody who's taken the covet vaccine in the world dies what happens to the world in five years from now or one year from now most people can't think because their brain just hijacks into amygdala more than oh my God everybody is dead I am dead and and they're like oh but they'll be like uh less retail market for this I was like but will there be Market will there be stock will there be money I think people cannot think about all of these things I think the variability you can create by putting like every movie with a thought experiment can be a great content so I think maybe we'll have some good content we will but I think I have a slightly different take this is so you spoke of unpredictability but this is predictable unpredictable I know that every time press that button there'll be some unpredictable thing it's guaranteed so but it's very Pablo even I think we'll keep pressing but you know hit what's it called hedonic adaption that's a thing yeah so we get bored of things you know we want that shiny phone in your hand right now you get it a week later it's become whatever then you need that other shiny thing that's how we keep buying and destroying the planet that's another thing but so if you and especially when it comes to art The Limited availability or whatever the exclusivity plays a key role a piece of painting by a certain person has immense amounts of value because it's that one particular person who made it it's not the aesthetic or visual quality it's all the other subjective parameters now if you're saying that there's a button that can generate art in any style infinite amounts of art it loses meaning right the I think the word is meaning so we're looking for meaning you look forward to a Nolan because Nolan's not making one movie every day then you wouldn't care about normal he's making that one amazing movie every X years and you're really looking forward if you have a Nolan button that keeps on generating you would you wouldn't want I know what you're saying I think maybe again a thought experiment just a while guess maybe people people will start preferring imperfect things that are human made maybe that will become more valuable so if you think about uh Japan as a society it's already living some of this already so what happens is uh there is trauma response and and it causes fight flight Fawn like different kind of possibilities and it creates certain behaviors in a human but a society can also go through a post-traumatic response so post that the bombings uh Japan has become fairly weird and I'm saying that word weird because from a human competitive lens it may not be weird from their own vantage point but uh like we could be looking weird right now to people who are listening in because we have a perspective I think definitely but the thing is that the Japanese found this interesting thing about like finding Beauty on such weird things right and and we'll be like why is that exciting and like like the like OCD is a trauma response and most people don't understand that like we're in this Obsession about pixelation and all of that and and it sounds morbid to say oh all the achievements are OCD driven and it is coming out of trauma which is again a dark version of how humans achieve anything in life uh is is like kind of playing out so I think what you're saying is finding the random things weird that means this is something unexplored I'm gonna enjoy this and enjoy this and so on and so forth is exciting I'm going to switch back to a different version of topic maybe we'll speak for a few minutes and then uh maybe go off camera and just have chat with the team uh because there were all these questions are about uh protocols like boring to ask you those questions that's really important in the face of planetary collapse important stuff so uh let's say you were forced not forced you were made the CEO CTO of India and uh assume that you are not existential and you're a very positive person and look forward to life and you want to save Humanity type Marvel feelings you have what are the three things three decisions you will take firstly I refuse to answer that because you're saying that I'm a whole different person my lived experience of let's say you know almost four decades is completely different my Outlook my knowledge everything is different I'm a whole different person I could I may also be I know Tom Cruise what would I do I don't know I can say anything it says it's an absurd premise I've always found those questions let's keep it let's say you are who you are and then I would definitely not want such a position yeah but is it because it's difficult is it because it's pointless is it because you don't think you can add value why would you say that so managing a tech team of let's say 30 people with very clear goals I mean finance and whatever that is not the same as running technology for entire civilization so it's just that it's Way Beyond me I can't even comprehend it and so incomprehensible that I wouldn't even dare to think or Daydream it's just I can't think you talk about this 30 plus people so I'm going to ask a specific question you have a a whole mutual fund product and let's say lending how many people run that I mean you know like two or three people and and that includes then there will be some product people and some design people and all of that all that's all in children it's all included but of course we have you know there are that's technology product ux UI Etc but there are people who do customer support Etc customer support teams that's not include but the entire product Tech team is for mutual funds that type pieces like three people okay so let's talk about the complexity that 30 people are running right now can you talk about the stack what's being managed by these 30 people so firstly I want to I want to add some clarity here it's not that 30 people can manage this it could have been then people it could have been 50 people could have been 100 people over a decade it has just evolved to be 30 people so that is not really a magic number could have been anything it just happened to be 30. it wouldn't be thousand though right no it would have a thousands of what that is so I mean stock broking is an extremely extremely complex business technology is complex the risks are infinite you could end up losing more money more Capital than the business has that's the nature of complex financial instruments uh sub millisecond latency all of those you know massive concurrency so there is just one layer of problems from a end user product perspective so we have to manage all of that and then the there's this whole invisible world of we call it EOD in the industry end of the day invisible world of countless complex processes that happen once the markets close we start getting massive data dumps from depositories exchanges you know clearing cops Etc and you have to take all of this crunch figure out the right accounts you know credit everyone's account etc etc etc so that's the entire back office process so the so these are this is just a product angle but then so zerotha has a thousand people we have a really large support team you know people call up asking about their money so we have to pick up the phone and answer we have legal we have compliance we have kyc we have to have all of that so to run an organization with a thousand people you need a lot of people systems crms intranet etc etc we manage all of those within the tech team we sell Force open source software our employee portal our support ticketing system our absolutely everything is in-house and self-hosted software so it's the same set of 30 people who run the front phasing technology back facing back-end technology the people technology organizational technology compliance technology Risk Everything but I'd like to say again that this has been an evolution it we've just we've built everything from the absolute Basics and we had the luxury of time we were not in a rush to do anything in two or three years so over a decade it's just happened to me fair enough uh um what is the well that's the worst possible on any vehicle could have yeah you feel important exactly I I sometimes feel that people honk because they feel so important because you gurus in many Societies in there like have much more worse traffic and companies honking almost disorienting for a bit and I don't know why people do that I think I don't have a very specific behavior and I'm sitting next to like some Caps or even sitting some friends like I sometimes saw somebody honking at a bumper and nobody's there like he just honked out of instinct like and I've known that TVs are now launching noise in the car I don't know if you know about this yeah yeah so that because pedestrians have not figured out to uh know that the car is coming unless there's a noise so we are putting artificial noise and enemies uh so we are special so I I'm gonna ask the last couple of questions uh in on a positive note uh uh what should we do to make some positive experiences now there's some ideas do you do anything positive in life everything I do is positive I am a realist it's very hard to once you see the evidence it's very hard to ignore the evidence and self sell yourself false optimism if the planet is facing an impacting catastrophe it's not something that you can unsee that whole willful ignorance thing I don't know how it works but like I said earlier to be human is to do positive things despite knowing that it might not even make a difference so everything I do I guess is positive all my hobby project software projects all my non software hobby projects the work that we do at our foundations the work we do at zero the the everything the stuff I do in personal life to do something what is classically defined as fun I mean I was listening to Beyonce just like I mean what everybody everybody else I can tell you one thing that uh I've known Mithun for now close to four five four five plus years and he's become a lot more like you or a period of time so I don't want to call it like he's become more pessimistic but more real so now when I'm meeting it's again dark conversations I used to have really nice conversations before uh okay last question um uh if we had to go back in time 50 years not not true one of the few things that and let's say you are now having the power to treat things one of the few things you would have tweet so that the humanity has we all love humans I'm assuming that we are still not rumors that go eight humans I should die you're not that we're not Thanos yet right so let's assume that we are positive so what would you treat it's again I mean even this concept of tweaking history given that billions of even shape random even shape everything that one tiny event that you may change make that you make 50 years ago could completely entirely change the world barring all of those paradoxes one thing that seems obvious would be to have a you know when globalization Etc was starting the world economy was being defined this whole thing about GDP production consumption maybe that should have been looked at differently while also so we we extract from nature infinitely where do we pay the price you sell me something uh I pay you there's that happening right you I can go to any shop by whatever I want you're paying the paying a price that factors into Infinite complexities in production but when people extract stuff from nature uh you know minerals uh Flora phone our trees whatever right you're not nobody's paying the cost and that's why we've ended up here right we've extracted unlimited oil unlimited water nobody's slid coins into the well it's piggy bank in every figurative sense so we should have had a different perspective of growth progress and economy in the technological era maybe 50 60 years ago where also took into consideration the natural wealth I think that would have been a very positive thing in general might have even curtailed the insistent consumerism might have maybe the global economies would have been much more sustainable and whatnot so perspective that seems obvious our idea of progress and development and growth should have been redefined to include the planet and the cost that we're supposed to pay for what we take so funny you're saying that I I was I always talk about this status driven society and wealth driven societies and one day it occurred to me easy to become a sustainable species to be statistical not wealth driven because wealth driven kind of creates this consumption machinery and wealth and all of that and all of that statistical Society is self-contained in some ways it's very mammalian and biologically relevant I don't think any species as well as driven every species is status driven and managed to survive and coexist with the constitution of environment but I think trying to become wealth driven is an interesting thing so on that note uh two fintech people are talking about wealth being a bad thing but that's what it is thank you okay thank you for doing this thank you thank you awesome thank you

Channel: CRED

Views: 86,022

Keywords: CRED, CRED curious, finance, payments, cred app, startups, kunal shah interview, Kailash Nadh Zerodha, Zerodha, Software Development, Software engineering, Nithin Kamath, Programming, MoneyControl, News, Coding, AI, Artificial Intelligence, AI Tools, AI Art, AI Robots, Open AI, Ai Voice, kailash nadh, jobs, policy

Id: 99J0tQOV5ks

Length: 62min 26sec (3746 seconds)

Published: Tue Jul 04 2023