Surviving the AI Age: Why India Must Go Back to the Factory Floor
A (non-satirical) Modest Proposal from a future economist for current policymakers
Indians have a cultural obsession with two types of jobs - government jobs because it allows one to achieve the FIRE (Financial Independence, Retire Early) lifestyle the moment they land the job, and ones in the IT Sector as it allows people to work in a nice office cubicle while retaining the possibility to achieve their FIRE goals. [Of course, the most efficient way to achieve the FIRE lifestyle is to launch a web10 start-up, use the VC money on ads, attract more VC money, and then shut it down to become a VC, but not before you write a twitter thread on how this was a huge learning experience]
The only reason why this obsession exists is because we are a status driven society, and we are a status driven society primarily because of scarcity. How do we get out of this scarcity? By investing in manufacturing.
Historically, countries that transitioned into manufacturing hubs—such as Japan, South Korea and China—experienced rapid GDP growth and employment expansion. India has the potential to follow a similar trajectory. World Bank reports suggest manufacturing contributed less than 13% to India’s GDP in 2023, while over 26% for China.
But why manufacturing now, when the world is going gaga over AI?
When I asked ChatGPT to give me an image for this article, it generated the one below using DALL-E:
While graphic artists and designers may have a hundred reasons to dislike it, it is good enough to displace those working in entry-level jobs. Sure, those working to design a new logo for JioHotstar will keep their jobs, but what about all the others?
People talk about how AI (read: LLMs) are gonna disrupt the IT sector, but it is also going to affect entry-level jobs in other sectors like copywriting, marketing and so on. So unlike what a lot of people used to imagine a couple of decades ago, it is the white-collar professions, and not the blue-collar ones, that are among the most at risk. It is not going to completely replace human beings of course, but one good employee can now easily do the tasks of ten ordinary ones, while earlier they might have been able to replace at most five of them. India, with its IT and back-office dominance, stands vulnerable.
However, I’m concerned that the discourse surrounding this issue is missing the forest for the trees. While there are certainly positive initiatives such as the IndianAI Mission and this other bit of encouraging news, the overall discourse seems to be on how India can upskill people to make them AI workforce ready. This to me seems like a defensive and reactionary response. Even if India does somehow manage to train the workforce based on current requirements, they might already be behind what is required in the industry once their training has been completed—remember, technological change grows exponentially (think about the time difference between inventing the wheel, the ox-cart, the bicycle, the car, the airplane, and the space shuttle).
Instead, we need to understand the crux of the problem—how do we deal with the emergence of AI (read: LLMs) so that two issues are sorted: first, that individuals have jobs, and second, that the economy continues to grow (and maybe the third issue of ensuring that this growth is equitable in nature). I do not think training the workforce to be AI-ready solves this problem in the long-run.
A Modest Proposal was a satirical essay by Jonathan Swift that suggested, with dark irony, that the poor in Ireland could solve their economic woes by selling their children as food for the rich. Swift's intent was to expose the cruelty of British policies toward the Irish. Now that AI models (read: LLMs) are improving at an almost exponential rate, here is a non-satirical modest proposal for India’s policymakers: instead of focusing on how to upskill people to prevent job losses due to AI, you need to focus on becoming the world’s premier manufacturing hub.
While work on this has already started, including in key industries like semi-conductors, we must push even harder while also selectively-ignoring the noise around AI.
Unlike in past industrial revolutions, AI has the potential to replace not just routine labor but also skilled knowledge work, shrinking employment opportunities in traditionally safe domains. To paraphrase something a professor of mine once said - Kapil Sibal doesn’t have to worry about LLMs, but junior lawyers and interns across the country whose job is to read case-files and prepare drafts may soon find that only a smaller percentage of them are now required. While AI will undoubtedly create new job categories, history shows that technological disruptions often result in transitional pain, with a lag before new sectors absorb displaced workers.
Additionally, India has a median age of around 28, significantly younger than countries like China and Japan who are close to the 40 and 50 year mark respectively. With over 500 million people in the labor force, the country possesses a massive demographic advantage that remains under-utilized. Unlike AI-susceptible sectors, manufacturing still relies heavily on human labor, particularly for assembly, quality control, and logistics. Moreover, India’s lower wage rates and relative geopolitical neutrality make it a prime candidate to absorb manufacturing shifts.
More importantly, one often overlooked factor is that AI adoption heavily favours English fluency, as most cutting-edge AI tools are designed, trained, and optimized in English. Based on personal experience, using highly precise language can give you 100x better results in LLMs, and maybe this differential might decrease as LLMs get better, but I do not think it will to a degree that one can get away with broken English. Unlike previous waves of technological change—such as IT outsourcing, which only required higher level managers to have high levels of English-fluency —AI does not inherently create new opportunities for non-English speakers at the same scale. In contrast, manufacturing, which relies less on English fluency, remains an accessible and scalable employment solution.
However, India’s logistics and infrastructure are still major bottlenecks. The World Bank’s Logistics Performance Index ranks India 38 as of 2023. Additionally, corruption at lower-levels of bureaucracy is extremely high, which could act as a heavy deterrent to individuals willing to establish their own factories. It is important to address this because only large existing players can otherwise thrive. While manufacturing will grow the pie, we also need to focus on evenly distributing the pie, which requires small players to enter the manufacturing game. Moreover, India should also streamline labor laws (so that we can stare at our wives instead of working for 70 hours a week) and improve its ease of doing business (I think we were somewhat 60th in the world by 2023) to encourage investments from abroad.
For those interested in solving the issue of corruption, check out this paper written by Banerjee, Baul, and Rosenblat which shows that individuals aspiring for bureaucratic positions exhibit higher levels of corrupt behaviour compared to private sector aspirants, although the likelihood of engaging in corruption remains similar across the two groups.
Manufacturing can also address broader structural issues beyond economic growth. Currently, too many students graduate with BA degrees from institutions that barely meet academic standards. Many enroll in these colleges under the belief that a degree guarantees success, only to find limited job opportunities, leading them to spend additional years preparing for government job exams. These degree mills contribute little beyond inflating education statistics—about as useful as boosting employment figures by hiring 500 people to dig a ditch with spoons instead of five with shovels.
To transform India into a dense manufacturing hub, vocational training and STEM education must take priority. Other countries have successfully implemented apprenticeship programs to bridge the skills gap, and India should adopt similar models at scale. However, a valid concern is whether India’s workforce, which is particularly accustomed to white-collar jobs, would accept this transition to manufacturing. This shift requires cultural and economic incentives, such as higher wages, better working conditions, and social campaigns that elevate the status of skilled industrial work—akin to how countries like Germany and Japan treat their technical workforce. The excessive glorification of the government job must end.
Finally, it is important to note that I am not suggesting we ignore AI, nor am I suggesting that we view it as an adversary. However, we must acknowledge realities about the short-term effects it can have over the next decade and how to best prepare for it. The rise of AI is inevitable, but India does not have to be a passive victim. By strategically pivoting towards manufacturing, the country can safeguard millions of jobs, capitalize on global supply chain shifts, and achieve long-term economic resilience. If India fails to act decisively, it risks falling into an unemployment crisis exacerbated by AI displacement.
Additionally, if you must invest in AI, perhaps better ways to do so would be to attract the best researchers in the world with huge amounts of funding, as well as manufacture in-house the technologies required for the same (work on this has already started). Do not focus merely on upskilling your workforce to use AI-tools. The best way to shield oneself from disruptions caused by technological change is to be the cause behind said disruptions. India would then be better prepared, while not only having the first-mover advantage, but also having more information as new technologies are created in-house.
Moreover, there also needs to be a discussion on whether India's AI strategy should focus on developing models trained in regional languages to make AI adoption more inclusive, thereby expanding the range of AI-driven jobs available to non-English speakers. However, one will undoubtedly encounter cultural and political constraints in this area, and hence, cannot proceed without solving the ‘language policy problem’ first. Researchers working on AI would have undoubtedly solved linear programming problems at some point in their lives, but India needs to solve a different kind of LPP before they can progress further.
It is also important to note that while manufacturing is a viable solution to AI-induced job displacement, automation in manufacturing itself is rising. To mitigate this, India should focus on labor-intensive industries before gradually moving up the value chain into high-tech manufacturing. This also buys time to adequately prepare our workforce for the new world we live in. Additionally, vocational training programs should emphasize a hybrid approach—teaching both traditional manufacturing skills and AI-driven automation techniques to future-proof workers.
A modest proposal, then: India must not wait for AI to disrupt its economy. It must counter it with industrialization, infrastructure, and innovation. Unfortunately, such changes will require great political willpower, and a great deal of ‘unselfishness’ from our bureaucracy. The next decade will define whether India ascends as a global manufacturing leader or struggles in the wake of AI-driven economic shocks. I hope short-term political benefits do not (once again) win over long-term economic benefits.