The AI factory: the rewiring of India's tech industry - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号):
请您阅读我们的用户注册协议隐私权保护政策,点击下方按钮即视为您接受。
FT商学院

The AI factory: the rewiring of India's tech industry

Outsourced data services may not be enough for India to thrive in the AI era
00:00

{"text":[[{"start":16.35,"text":"I think the jury is out there about what is AI. "}],[{"start":19.700000000000003,"text":"AI is fundamentally a marketing term. It has become a way to unlock vast amounts of capital. "}],[{"start":25.650000000000002,"text":"Well, $1.5tn is being pumped into these large language models just to keep them learning. "}],[{"start":31.700000000000003,"text":"With automation, 60% of the work is automated. I don't need 100 people. I need only 40 people. "}],[{"start":38.95,"text":"I think before we talk about what is India's unique AI path, maybe we need to discuss: does India have a unique AI path? I mean, that's what India wants to tell the world. "}],[{"start":50.25,"text":"We are probably the second-largest AI workforce in the world today. "}],[{"start":54.95,"text":"Any profession where the task is well defined, those tasks will be done by AI. "}],[{"start":60.25,"text":"While certain kinds of tech jobs may go out, there will be certain other newer kinds of tech jobs which will get created. "}],[{"start":67.05,"text":"AI should be understood as a technology that, by design, concentrates power. "}],[{"start":72.6,"text":"The resources that are being generated in the West and in China, India just doesn't have that. India is still largely the back office for the world. What will be the future of AI in India? What will it be used for? Is it exploitative? Are the people aware they might be creating data that will lead to robots replacing them? "}],[{"start":91.35,"text":"Karur is a very small town in Tamil Nadu, a southern state, and you don't have a lot of jobs in these places. We visited this textile manufacturing factory in Karur. There was a station where they are stitching something, there was a station where they are folding something, there's a station where they're ironing something. And almost in all of these stations, some of the workers were wearing cameras on their foreheads. It was either a GoPro, and some of them were also wearing Meta Glasses. "}],[{"start":119.55,"text":"We have 200 people, and we utilise at least 60 people for this robotics area. They wear it every day, at least for 6 to 8 hours per day. They have the additional benefit of 10,000 rupees per month. "}],[{"start":133.7,"text":"It looks very dystopian to start with. Is this what humans will become? Are we training robots now? Are we the fodder for robots? I mean, it doesn't seem like a tech job, but if this camera will tomorrow train how Tesla produces a new robot, that's a tech job. But this is the bottom end of the tech job, where you are not the person who's actually producing any technology, but what you're doing is still training technology. "}],[{"start":159.95,"text":"Objectways started in 2019. We basically started as a data annotation company. Later, we evolved into robotics and we are, right now, collecting all types of data, egocentric data, gripper data, daily operations data. All sorts of data has been collected for robotics. Basically, the egocentric data is nothing but all the day-to-day activities which a normal human being performs. "}],[{"start":183.6,"text":"You have to understand, for a robot to know that this is a cup, San Francisco offices don't train the robot to do that. It's people sitting in small-town India. "}],[{"start":193.79999999999998,"text":"The fact that we have 1.5bn people, who can generate data, is both good and bad. "}],[{"start":200.7,"text":"This kind of data work, to me, provokes that fundamental question, which is, this is a supply chain which is extractive by design. "}],[{"start":208.64999999999998,"text":"If you're paying a price and somebody is willing to do it, then it's a fair trade. But at some point, somebody has to ask, are we being exploited? Is there an alternative? "}],[{"start":220.89999999999998,"text":"Are we data working our ways out of employment, out of good jobs, into a world where the mediating force, the technology that mediates across different sectors of the economy are the companies that run away with the lion's share of the value? "}],[{"start":235.24999999999997,"text":"There are geographies in India where people will be exploited. They will be asked to wear these cameras by their employers with no extra money given to them. And all this data can be used against the workers as well. "}],[{"start":248.39999999999998,"text":"The creation of these large language models has raised really serious concerns about whose knowledge was it that created these systems that these companies are now profiting in the billions out of? "}],[{"start":258.59999999999997,"text":"India's saying, we'll build our own large language models, we'll build our own products. But almost all the products that are there globally at some point will be trained by somebody in India. India has had a massive data annotation industry for many years now. What is data annotation? It's basically, you tell every piece of information, you explain what that information is for the computer to learn. There are self-driving cars in the world. Those cars need to identify what are the objects on the street. "}],[{"start":291.15,"text":"This is a truck. This is a car. This is a billboard. You annotate all that and train the algorithm. It goes back to about 15 years. NextWealth started when we went to a place called Mallasamudram, and we saw this huge engineering college there with thousands of graduates coming out of the college. We went back and looked at the data, and we realised that 60% of the colleges are concentrated in the small towns. Therefore, 60 per cent of the graduates in the country are also from small towns. "}],[{"start":322.29999999999995,"text":"And these are first-generation graduates, which means parents are farmers, daily-wage labourers, agriculturists, tailors, clerks, policemen and they would have taken loans for their education. "}],[{"start":334.54999999999995,"text":"We work with some of the top companies in the retail segment. They are doing absolutely cutting-edge AI work in the space that they are in. When they are building these AI algorithms, they need people who are able to train, test, and validate these AI algorithms, which means data annotation, video annotation, audio annotation, multi-turn conversations, reinforced learning through human feedback. "}],[{"start":358.15,"text":"If there's an assembly line and one person's job is to put a screw in, that person has to make sure the screw is going in accurately every time for every product that passes through him or her. It's the same for data annotation. It's like a factory floor. You have to have absolute accuracy. "}],[{"start":374.9,"text":"For 1.5bn people that live in India, you have three large cities that are the main growth drivers, Delhi, Mumbai, and Bengaluru. Delhi and Mumbai have a population of about 20mn, 22mn. Bangaluru has about 15mn. But if you have to create jobs, not everybody can be here. "}],[{"start":397,"text":"In a country like India, which has been a socially conservative nation, where women in many parts of the country are still not even allowed to be properly educated, if you can give women jobs where they don't have to leave their hometowns, if it's data annotation work, that's immensely adding value to their lives. "}],[{"start":414.2,"text":"There is a saying. If you educate a boy, you educate the family. If you educate a girl, you educate the entire village. "}],[{"start":421.55,"text":"So for people here from the small towns, while their aspiration is to work in these IT and BPO organisations, they are not coming to recruit here. So the local job opportunities are very limited. So we said, why don't we match the demand and supply and take the work to where the people are, rather than people travelling to where the work is? "}],[{"start":440.7,"text":"So when I think about AI in India, India has a lot of ambitions for AI, but it doesn't really have either the resources, or it hasn't put in the effort to really take the lead in any of these new categories that AI is producing today. India is still thinking about what are the fastest or maybe the cheapest route to find lead in one of these races, but it's not found that yet. "}],[{"start":461.84999999999997,"text":"I wouldn't agree entirely with that characterisation. I think there are many reports which indicate we are in the third place as far as the global AI race is concerned. Eventually, I think the impact that AI will have on economies is, basically, in terms, the extent of adoption of AI within the economy. And I think you will find, ultimately, that India will turn out to be a winner in that space. "}],[{"start":483.54999999999995,"text":"We are sitting in Bangalore right now. This city was built on the tech industry, the boom of the tech industry that started in the early '90s in India. At the moment, India services, they contribute about $330bn to $340bn in exports, which is larger than any one single goods exports for this country. That's extremely important for India's economy because that's a major source of dollar income. That's also the basis of the largest export that India has. And that's the information technology services. "}],[{"start":516.6999999999999,"text":"The availability of talent, the capability of these engineers who are coming out of these premium institutions, some of them already worked for a few years here and there. Very highly qualified people who understand systems well, who could quickly learn what is there in the systems. One, of course, English language was a huge advantage. India has served as the back office for the multinational giants for more than 25 years now. Under promise, over deliver. That actually probably sealed it for the industry. "}],[{"start":545.65,"text":"The core of what India's tech industry sold to the world was: we will solve all the tech problems for you, whether it's routine stuff, like HR or accounting or some softwares that you require. "}],[{"start":558.6,"text":"The fact that we were at the right price point was a huge deal as well. "}],[{"start":563.1,"text":"If you are paying $1 to a person for a particular job, I can get it done for you for 20 cents. AI is probably the toughest challenge that India's IT services sector has faced in their existence. It is absolutely threatening the dominance that they had for the last 25 years. "}],[{"start":581.1,"text":"A lot of the work that is currently being done by the Indian IT service industry, which would mean that there not only is there exposure, but their risk of their jobs being fully displaced rather than simply transformed is really, really high right now. "}],[{"start":596.95,"text":"The kind of job loss that AI is going to bring about in white collar is going to be significant. But there is enough anecdotal evidence from companies that the rules have changed, that the game has changed. "}],[{"start":612.2,"text":"This current AI moment is bringing that very same kind of threat to the Indian IT services industry that was born out of that moment of neoliberalism. "}],[{"start":621.5500000000001,"text":"My name is Krishn Kaushik. I am the Mumbai correspondent for the Financial Times. I've been reporting on India for almost 15 years now. After liberalisation, when money started to trickle into the country, you started seeing these new products that you just hadn't thought about earlier, because India had always been so inward-looking. You could eat a Maggi, which is instant noodles. You would get Pepsi, Coke. "}],[{"start":646.0000000000001,"text":"For me as a kid, initially, just aspiring for a Nike shoe was a big deal because you couldn't find Nike shoes, and Nike shoes were extremely expensive compared to what you would get domestically. So these international brands seeped into your everyday life. "}],[{"start":661.3000000000001,"text":"By and large, the IT industry as we know it has changed forever. So India cannot be complacent. "}],[{"start":667.95,"text":"You don't have to hire software engineers to create these softwares to make your businesses more streamlined. Now, a lot of those businesses can just do that using LLMs. "}],[{"start":678.4000000000001,"text":"The kind of high volume, repetitive tasks, but also the margin of error that might be somewhat acceptable in a large majority of the contexts in which the IT services industry is working make it a prime candidate for AI substitutability. "}],[{"start":691.2,"text":"The sector's dominance is surely threatened because they were not the vanguard of bringing AI to India. That was their problem. They were immensely successful, immensely profitable. They were not investing in AI. "}],[{"start":704.0500000000001,"text":"Where we have really lagged is private sector investment in R&D. I think part of it reflects a lack of faith in some of the companies in their own ability to make these investments and make sure that these become paying investments, because the risks are larger. "}],[{"start":719.0500000000001,"text":"Tesco is one of the oldest retailers in Europe. It is a £66bn business. We are not just a retailer, we are a retailer, wholesaler. 100% of all architectural design of Tesco is done from Tesco Business Solutions and right here in Bangalore. Whether it's the civil engineering, whether it's mechanical engineering, electrical engineering, we have over 100,000 fridges across our 4,500 stores. And these are managed through an IoT platform here where we monitor fridge temperatures. "}],[{"start":750.4000000000001,"text":"If I look at the most complex thing that we do is something called personalisation. We have data over the last 30 years for around 22mn customers. And what AI does is really understand the pattern, the trend, and really help our customers to shop better. We've used AI in a tool called Cost Intelligence Model. A bread company came to us to say: we're going to increase the cost of bread by 10%. "}],[{"start":776.9000000000001,"text":"So what the AI tool did was break up the bread ingredients to say: it's made up of yeast, it's made up of water, it's made up of flour. And therefore, what's inflating and what's deflating? So that helped us to reduce our costs. "}],[{"start":791.0500000000001,"text":"India's still doing the work for companies that are largely based outside the country. That's why it's called the back office. Now, a lot of that work is being done in-house in what are called global capability centres. "}],[{"start":803.5500000000001,"text":"Today, if a store has any issue, we have an AI system which senses that issue, categorises that issue into one of 300 possible problems that could happen in the store. And then send the right vendor into the store to fix it as fast as we can. All of this happens touchless. "}],[{"start":820.95,"text":"I think people talk about job losses. And what AI would do. I think of AI slightly differently. A lot of people said when internet would come in, it would create a lot of job losses. But it doesn't. The threat to humanity is not AI. Threat to humanity is the inability to learn. "}],[{"start":840.1500000000001,"text":"When I think of AI in India, the first word that jumps to my mind is a scramble. India is trying to project everything. India is saying saying, we'll build our own applications, we'll find our own solutions. But none of that is visible at the moment. "}],[{"start":855.7,"text":"By virtue of India's IT skills, I believe India can play a very important and unique role in AI. "}],[{"start":862.45,"text":"I think India actually has the opportunity to be the AI factory of the world. The booming AI industry needs more and more humans in the loop. "}],[{"start":871.7,"text":"I think there's a distinct possibility that the overall number of jobs in the tech sector in India will actually go up and not go down. We could become a location from which AI-trained talent can be provided for the rest of the world. "}],[{"start":884.45,"text":"India has a population of 1.5bn. India wants to leverage the cheap workforce that's available over here to find its spot in the global AI race. "}],[{"start":894.8000000000001,"text":"I think offering up the scale of our population as a carrot to attract foreign tech companies is not a pathway to anything resembling sovereignty or resilience longer term in the AI age. "}],[{"start":906.35,"text":"If you just go by the markets, investors are fleeing Indian markets at the fastest pace ever. I think just in the first three months or first four months of this year, some $25bn worth of investment has left the country. They are moving their money to markets like Taiwan and South Korea. Because there, those countries are either producing chips or they're producing some other hardware that's relevant for this AI race globally. India doesn't have that. "}],[{"start":932.8000000000001,"text":"AI is almost a symbol for all of the ways in which the economy feels rigged to the everyday person. "}],[{"start":940.3000000000001,"text":"India will become the use-case capital of the world with the way things are going. But that does not mean we should do it at the cost of sovereignty. "}],[{"start":948.2,"text":"I think this kind of framing that positions us as a market for the taking for US tech giants, in some ways, cultivates a position for India in the supply chain that is more akin to the back room with echoes to the BPO era. "}],[{"start":962.0500000000001,"text":"This is a young country. There are people who are willing to experiment with what comes their way. There are people who are willing to adapt and change things quickly. If India is to become a developed country by 2047, then this is a technology which we need to adopt and adapt to our circumstances. We need to make sure that this is a technology that works in all the real sectors of the economy. "}],[{"start":985.45,"text":"India has one unique thing which almost nobody has. That's talent at a very, very large scale. Now, the challenge over there is, if you don't have enough jobs for these people, then what is the value of this scale? It's nothing. "}],[{"start":1008.1,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1782545430_2330.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。
设置字号×
最小
较小
默认
较大
最大
分享×