AI Humara Job Kha Jayega? — To Train Our Own Foundation Models
A short history of automation panic and an argument for learning to build/train AI rather than only fearing it.
One Consolidated History: From Calculators to Computers, ATMs, and the Internet in India
“AI humara job kha jayega” is not a new sentence in Indian social life. It is the latest form of an old reflex that appears whenever a technology moves from assisting a worker to performing part of the worker’s role. If the word “AI” is replaced by “calculator,” “computer,” “ATM,” “internet,” or “smartphone,” nearly the same anxiety can be found in earlier periods, usually voiced by incumbent workers just before younger workers adopt the tool and rebuild occupations around it (Bessen, 2015; Visier, 2022). This first section consolidates the non-AI history into one narrative: how these technologies originated globally, how they arrived in India, what fear stories emerged in Indian society, and how adoption transformed work rather than simply eliminating it. The purpose is not to romanticise technology, but to establish a factual pattern before examining why artificial intelligence is both continuous with that pattern and different from it.
Globally, the story begins long before modern computers. Blaise Pascal built the Pascaline in 1642 at the age of eighteen, partly to reduce the arithmetic burden on his father, a tax official, and Gottfried Leibniz later contributed to mechanical calculation by adding automatic multiplication after 1671 (Wikipedia contributors, n.d.-c). Yet for nearly three centuries, the word “computer” did not primarily mean a machine. It meant a person, often a woman, who performed calculations manually for astronomy, banking, engineering, insurance, and eventually spaceflight; NASA employed many human “computers” before electronic machines took over that function (National Women’s History Museum, n.d.). The electronic computer of the twentieth century retired that job title, but it created an entirely new industry around programming, systems analysis, data management, and software engineering. After that came office automation: personal computers, word processors, spreadsheets, and accounting software reduced the cost of storing, copying, calculating, and retrieving information. Banking automation followed through the automated teller machine, first deployed in the late 1960s, which allowed routine financial transactions to occur without a human teller performing every step (Wikipedia contributors, n.d.-a; Bessen, 2015). The internet and later mobile platforms then reduced the cost of communication, search, distribution, payment, logistics, and coordination, shifting entire economic activities from physical intermediation to digital systems.
India’s encounter with these technologies was layered rather than sudden. The country’s earliest computing presence was institutional and scientific, appearing in statistical, research, and educational institutions in the 1950s and 1960s, long before computers entered ordinary offices (Wikipedia contributors, n.d.-b). The mass arrival happened later, especially during the 1980s and 1990s, when personal computers began entering Indian offices, coaching institutes, chartered accountants’ firms, banks, trading houses, and businesses. This period coincided with economic liberalisation, when Indian firms began competing with global management practices and digital record-keeping became commercially necessary. Desktop accounting software, including Tally and similar packages, became part of the first major wave of Indian business software, changing how small and medium enterprises maintained accounts (LSVP, 2014). Banking automation arrived through ATMs, which entered Indian banking in the late 1980s and expanded rapidly after the 2000s as private banks competed on convenience, branch reach, and customer service (EPW Research Foundation, n.d.; Reserve Bank of India, 2013). The internet and smartphone wave followed a similar path: from elite curiosity to mass infrastructure. Dial-up internet in the 1990s gave way to broadband, mobile internet, cheap data, smartphones, digital payments, and platform businesses. The Unified Payments Interface, developed by the National Payments Corporation of India, became one of the most important examples of India adapting digital infrastructure to its own scale and public-interest needs (National Payments Corporation of India, n.d.).
Each arrival produced a fear narrative. When calculators became common, the fear among clerks, accountants, and bookkeepers was effectively that arithmetic skill would become useless: “calculator aayega, hisab khatam ho jayega.” When personal computers entered offices, the fear expanded to typists, clerks, and file-keepers: “computer aayega, babu khatam ho jayega.” When accounting software spread, traditional accountants feared that software would replace professional judgement: “software aayega, accountant ki zaroorat khatam ho jayegi.” When ATMs appeared, bank employees and labour voices feared that machines would remove front-line banking jobs: “machine aayegi, clerk ki naukri chali jayegi.” When the internet, e-commerce, and smartphones grew, traders, transport workers, media persons, and service providers feared that digital platforms would destroy traditional livelihoods: “online aayega, dukaan khatam ho jayega.” These phrases are not presented here as archival quotations unless specifically cited. They represent the recurring social grammar of technological fear in India: the worry that a new machine will not merely assist the worker but invalidate the worker’s accumulated skill, status, and livelihood.
Despite the fear, adoption happened, often slowly at first and then quickly among the young. The calculator did not end accounting; it changed accounting. Once arithmetic became cheap, the value of the accountant shifted from calculation to interpretation: auditing, taxation, compliance, financial planning, and advisory work. The computer did not end offices; it digitised them. New roles emerged, including computer operator, data-entry operator, Tally accountant, office automation executive, IT support engineer, programmer, software tester, systems administrator, and business process outsourcing associate. India eventually built a large technology-enabled services economy around these skills. The ATM story is the most misunderstood automation case. In the United States, the most studied version of this paradox, ATM numbers rose from about 60,000 in 1985 to 352,000 in 2002, while teller employment rose from about 485,000 to 527,000 (Visier, 2022). Economist James Bessen explained the mechanism: ATMs reduced the number of tellers needed per branch, making branches cheaper to open, so banks opened more branches and the surviving teller’s role shifted from cash handling toward customer service, sales, relationship banking, and problem-solving (Bessen, 2015). However, the ATM story also contains a warning. It worked as a job-expanding story because the ATM automated a task, not the entire role. When automation becomes near-complete, the effect can change. Commentator Jacob Morgan, summarising more recent firm-level research, notes that employment tends to fall by roughly 14% once a technology crosses from partial task automation to near-complete task automation (Morgan, 2026). In India, mobile banking and UPI moved financial transactions closer to near-complete automation, altering the meaning of front-line banking work more deeply than the ATM alone had. The internet and smartphone economy similarly destroyed some intermediaries but created many new job families: web developer, app designer, UI/UX researcher, digital marketer, SEO executive, social media manager, product manager, growth analyst, delivery operations manager, logistics coordinator, fintech operations associate, payment specialist, content creator, and gig-platform manager. None of these roles existed in their current form in 1995. The consolidated lesson from non-AI history is that these technologies did not merely take jobs. They unbundled jobs into tasks, automated routine tasks, and pushed humans toward judgement, coordination, sales, analysis, design, and decision-making. They moved work, imperfectly and unevenly, from “hard work” to “think work.” Artificial intelligence now asks whether that pattern can survive a machine that can also draft, code, summarise, analyse, reason, and act autonomously.
AI’s Global Origin, Arrival in India, Autonomous Agents, and Reasoning Breakthroughs
Artificial intelligence is different in kind from the earlier technologies described above. Traditional software was built for one purpose: calculate interest, store records, dispense cash, route a call, or display a webpage. A foundation model is trained broadly on large amounts of text, code, images, or multimodal data and then adapted to many purposes. It can write an email, debug code, summarise a judgment, translate a document, explain a scientific concept, generate an image, or draft a business plan from the same underlying architecture. This generality explains its unusual adoption speed. ChatGPT reached approximately 100 million users in roughly two months after its late-2022 launch, a pace TikTok took nine months to match and Instagram took two and a half years to match (Bhaimiya, 2023). No earlier consumer-facing automation technology had spread that fast.
India encountered AI through multiple channels simultaneously: IT services companies embedding AI into software delivery, startups using AI for coding and product development, fintech firms using AI for fraud detection and credit assessment, edtech platforms using AI for personalised learning, healthtech companies using AI for diagnostics support, creators using AI for scripts and translations, students using AI for exam preparation, and professionals using AI for reports and analysis. AI in India is therefore not only a research topic; it is already a workplace instrument. NASSCOM reports that 86% of employers have seen some impact of AI on job roles and responsibilities, with 35% witnessing significant redefinition or transformation (NASSCOM, 2026). The NITI Aayog, Boston Consulting Group, and NASSCOM roadmap further states that AI disruption is already reshaping jobs in India’s technology and customer-experience sectors (NITI Aayog et al., 2025).
The most important technical difference is autonomy. Current autonomous AI systems do not operate in a philosophical vacuum; a human still defines the objective, constraints, evaluation metric, and safety boundary. But within that frame, the agent can plan, generate, test, revise, and improve outputs without continuous human intervention. Google DeepMind’s AlphaEvolve is a clear example: it is described as an autonomous pipeline of large language models that generates candidate algorithms, tests them, and evolves better ones through a feedback loop, while the human role is to define what “better” means before the loop begins (Novikov et al., 2025). Google has reported using AlphaEvolve to improve data-centre scheduling, assist chip design optimisation, and accelerate training processes for the very language models that power the system itself (Google DeepMind, 2025a). In mathematics, the change is equally striking. In 2023, no AI system could score even a single point at the International Mathematical Olympiad. In July 2024, Google DeepMind’s AlphaProof and AlphaGeometry 2 solved four of six problems, achieving a silver-medal standard (Google DeepMind, 2024). By July 2025, Google DeepMind reported that an advanced version of Gemini with Deep Think achieved gold-medal standard, working within the same time limits and constraints given to human contestants (Google DeepMind, 2025b). OpenAI also reported gold-medal-level performance in 2025, though with the important distinction that its result was graded by former medalists on its own terms rather than through the official IMO certification process (OpenAI, 2025; Willison, 2025). In the physical sciences and engineering, AI is also contributing to discovery, including large-scale materials prediction and optimisation problems where physical constraints matter (Merchant et al., 2023; Novikov et al., 2025). The careful claim is not that AI has replaced mathematicians or physicists, but that AI is becoming a reasoning partner that can explore solution spaces, propose candidates, and accelerate discovery under human direction.
The Indian Fear Narrative: “AI Humara Job Kha Jayega”
The AI fear in India is sharper than in many countries because Indian employment is highly aspirational, exam-driven, and credential-sensitive. Students fear that AI will devalue degrees. IT professionals fear that coding assistants will reduce demand for junior developers. Chartered accountants fear that AI will automate auditing and tax work. Lawyers fear contract review automation. Doctors fear diagnostic AI. Creators fear generative images and video. BPO workers fear voice and chat automation. Teachers fear that personalised AI tutors will undermine classroom authority. Parents fear that the old ladder of degree, job, promotion, and security may break beneath their children’s feet. A major Indian policy document captures this mood directly: the NITI Aayog, Boston Consulting Group, and NASSCOM roadmap states that automation has threatened nearly 60 percent of jobs in India and that Indians are more afraid of AI taking away jobs than their global peers (NITI Aayog et al., 2025).
This fear is not irrational, but it is often imprecise. The global precedent is instructive. In 2013, Oxford researchers Carl Benedikt Frey and Michael Osborne published the most-cited automation statistic of the decade: 47% of US jobs were at high risk of computerisation within 10 to 20 years (Frey & Osborne, 2013). The figure became gospel in boardrooms and headlines. Nine years later, the Information Technology and Innovation Foundation checked the record: the US had added 16 million jobs since the study, unemployment stood at 3.7%, and the correlation between an occupation’s predicted “automation risk” and its actual job losses was a weak 0.26. Insurance underwriters, rated among the highest-risk occupations, saw employment grow by 16.4% instead of vanishing (Information Technology and Innovation Foundation, 2022). The lesson is not that automation never happens. The lesson is that predicting which jobs disappear, which transform, and which emerge is far harder than one headline number suggests.
The World Economic Forum’s Future of Jobs Report 2025 provides the most rigorous current forecast. Based on a survey of over 1,000 employers across 55 economies, it projects that AI and related technology trends will generate 170 million new jobs globally by 2030 while displacing 92 million others, producing a net gain of 78 million jobs (World Economic Forum, 2025). But the same report carries two honest warnings: 40% of employers expect to reduce their workforce where AI can automate tasks, and 39% of the core skills workers use today will be transformed or outdated within five years. This is the Indian dilemma in miniature. The fear narrative says AI will eat the job. The evidence says something more complicated: AI may eat the task, reshape the role, eliminate some occupations, create new ones, and increase the premium on judgement, verification, domain expertise, and human direction. The real anxiety in India is not only unemployment. It is also degree devaluation, loss of first jobs for freshers, automation of junior-level work, pressure to upskill faster than institutions can teach, fear that AI may benefit only elite English-speaking technologists, concern about dependence on foreign models trained on non-Indian data, and anxiety that public institutions will move slower than private platforms. Thus, “AI humara job kha jayega” is not merely an economic statement. It is a social fear about status, mobility, language, education, and national capability.
Adoption, New Roles, Education, and India’s Foundation-Model Mission
If the historical pattern holds, the youngest workers should adopt fastest, and they are. A 2025 survey reported that 82% of Gen Z employees use AI at work, compared with 52% of Baby Boomers, while another survey found that 70% of Gen Z workers taught themselves AI tools rather than waiting for formal training (CFO Brew, 2025; FM Magazine, 2025). India mirrors this pattern with its own intensity. NASSCOM projects that India’s AI talent pool is expected to grow from around 600,000–650,000 to over 1.25 million by 2027, and it reports that India recorded close to 100% year-on-year growth in prompt-engineering talent in 2025, with AI engineering talent hiring rates placing India ahead in parts of the global technology workforce curve (NASSCOM, n.d.-a; NASSCOM, n.d.-b). Young Indians are not merely consuming AI. They are using it to learn coding faster, prepare for competitive exams, build startups with smaller teams, create content in Hindi, English, and regional languages, automate freelance workflows, analyse data without being formal data scientists, draft legal, financial, and medical documents under supervision, and build portfolios without traditional gatekeepers.
This adoption is producing new job families. Emerging roles include prompt engineer, AI trainer, data annotation specialist, reinforcement learning from human feedback specialist, AI product manager, AI solution architect, MLOps engineer, LLMOps engineer, AI ethics officer, model evaluation analyst, Indic language data curator, sovereign AI researcher, AI compliance analyst, automation consultant, AI-enabled customer experience designer, AI safety evaluator, synthetic data engineer, retrieval-augmented generation specialist, and AI operations manager. These roles are not speculative job-title invention. They reflect the labour churn described by the World Economic Forum, NASSCOM, and the NITI Aayog roadmap: old tasks are being automated, while new tasks around building, evaluating, governing, localising, and directing AI are emerging (NITI Aayog et al., 2025; NASSCOM, 2026; World Economic Forum, 2025). The important shift is that the worker who only executes becomes vulnerable, while the worker who can direct, verify, contextualise, and build becomes more valuable.
India’s education system is not yet fully aligned with AI, but policy and skilling structures are beginning to move. The National Education Policy 2020 emphasises foundational literacy, digital literacy, coding, computational thinking, and exposure to emerging technologies, which is important because AI literacy cannot remain only a postgraduate specialisation (Ministry of Education, 2020). At the higher-education and skilling level, institutions and platforms are expanding AI-related learning through IITs and other technical universities, NPTEL and SWAYAM-style digital course platforms, NASSCOM FutureSkills and industry-aligned training, startup incubators, corporate reskilling in IT services and banking, financial services, and insurance sectors, and government-backed AI education initiatives. NASSCOM’s skilling analyses stress that India needs interventions to deepen AI capability, reduce skill gaps, and convert talent volume into productive AI expertise (NASSCOM, n.d.-a). But adding “AI” to a syllabus is not enough. The deeper challenge is pedagogical: Indian education must shift from memorisation to problem-framing, from answer reproduction to verification, from individual exam performance to human-AI collaboration, and from rote compliance to ethical judgement.
The most important Indian response to AI fear is not protest but construction. India is trying to move from being a consumer of foreign AI systems to becoming a builder of sovereign, Indic, and culturally grounded foundation models. The IndiaAI Mission represents a major national push to develop domestic AI capability, including compute infrastructure, models, datasets, and talent; research on India’s developmental AI approach notes that the mission allocated a large share of its programme budget, roughly USD 600 million, to AI compute infrastructure (Sharma, 2025). This matters because foundation models require compute, data, and talent, and India has talent at scale, growing data diversity, and now a policy attempt to secure compute and sovereign model capability. In April 2025, the Government of India selected Sarvam AI under the IndiaAI Mission to build India’s sovereign large language model, with the stated ambition of an indigenous model trained and hosted within India, attentive to Indian languages, cultural context, and data sovereignty (Sarvam AI, 2025; Government of India, 2026b). Concurrently, the IIT Bombay-led BharatGen initiative has developed Param-1, a bilingual foundation model built from scratch for English and Hindi, described as a 2.9-billion-parameter model trained with substantial Indic-language representation, addressing the linguistic gap in many global models (BharatGen, n.d.; CXO Digital Pulse, 2025). In 2026, the Government of India highlighted the launch of BharatGen’s Param-2 text foundation model supporting 22 scheduled Indian languages (Government of India, 2026a). This is not merely a technical achievement. It is a civilisational data project, because a language model trained mostly on English internet data does not naturally understand Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Odia, Punjabi, or the mixed code-switching reality of Indian speech. India’s foundation-model ambition is also enabled by the global open-source ecosystem. Hugging Face hosts more than 500,000 open AI models, datasets, and tools, often described as the “GitHub for AI,” lowering the entry barrier for Indian researchers, startups, students, and public institutions (AI Solutions by Davies Meyer, 2026). The strategic lesson is clear: India should not only ask whether AI will take Indian jobs. India should ask whether Indians will train, evaluate, govern, and localise the models that shape those jobs.
AI as Assistant, Not Job-Eater
The Indian narrative often swings between two extremes: “AI will take all jobs” and “AI is just hype; nothing will change.” Both are wrong. The evidence supports a third position: AI will automate tasks, transform roles, create new job families, and increase the premium on judgement, verification, domain expertise, and human direction. The World Economic Forum projects net job creation globally by 2030, but also warns of large-scale skill disruption (World Economic Forum, 2025). NASSCOM finds that employers are already experiencing role redefinition rather than simple elimination (NASSCOM, 2026). The NITI Aayog roadmap highlights both fear and opportunity, showing that India’s labour market is anxious but also positioned to capture AI-led employment if skilling and innovation keep pace (NITI Aayog et al., 2025). The ATM lesson remains relevant, but with a caution: partial task automation often creates new roles by lowering costs and expanding demand, while near-complete task automation can force deeper role redesign (Bessen, 2015; Morgan, 2026). AI is dangerous precisely because it is not limited to one task. It can draft, summarise, code, analyse, translate, and reason across many tasks. Yet even here, the human role does not become meaningless. It becomes different. The AI can draft the contract; the lawyer must judge risk, precedent, client intention, and ethics. The AI can generate code; the engineer must architect systems, verify security, and understand business constraints. The AI can summarise research; the scientist must choose the question, interpret the meaning, and design the next experiment. The AI can solve an olympiad problem; the mathematician must decide which problems matter. The real risk is not AI alone. The real risk is unequal AI capability. A worker who can use AI to research, draft, code, analyse, and verify becomes more productive. A firm that builds AI workflows becomes faster. A country that trains its own models becomes less dependent. A student who learns to direct AI becomes more employable. A citizen who cannot access AI capability may be left behind. Thus, the answer to “AI humara job kha jayega?” is not a simple yes or no. AI will eat the routine. AI will amplify the strategic. AI will reward the builder. AI will punish the passive.
Conclusion: From Fear to Capability
The story from Pascal’s calculator to India’s sovereign foundation models is not a story of machines defeating humans. It is a story of humans repeatedly outsourcing routine effort and then being forced, and often empowered, to move toward judgement, design, relationship, ethics, and direction. The calculator took arithmetic. The computer took records. The ATM took cash handling. The internet took distribution. AI is now taking parts of drafting, coding, summarising, analysing, and even iterative reasoning. But each wave created new work, not the same work, not always in the same place, and not without pain, yet new work nonetheless. India’s current advantage is demographic and digital: a young population, a large software workforce, expanding internet penetration, a startup culture, and a government that is now trying to build sovereign AI capacity through missions, startups, academia, and open ecosystems. The decisive question is not whether AI will take a job. The decisive question is whether individuals, firms, universities, and the nation will learn to direct, verify, localise, and build AI before it is directed, verified, localised, and built without them. If India answers that question with capability, then the old fear, “AI humara job kha jayega,” may become another historical echo rather than a prophecy.
Acknowledgement / Methodology: How This Research Story Was Built
This article was developed through a human-AI collaboration that mirrors its own argument. The human author and director, Subhadip Datta, set the central question, the title, the Indian sociological frame, the narrative arc, and the final editorial judgement. He asked for a fact-based story rather than a generic blog, insisted on the sequence of global origin, arrival in India, fear narrative, adoption and transformation, and required that the AI section be expanded with evidence on education, sovereign foundation models, autonomous agents, mathematics, physics, and AI as assistant rather than job-eater. Claude, Anthropic’s AI model, assisted in the earlier research and drafting phase, including initial web-based source gathering, structure proposal, and fact-checking orientation. Qwen, Alibaba’s AI model, then helped restructure the piece into the final social-research narrative, deepen the Indian context, verify source alignment, expand the AI and foundation-model sections, and refine the APA citation framework. The process was not “AI writes, human copies.” It was closer to the future of work this article describes: the human defined the question and the moral and intellectual direction; the AI models accelerated literature retrieval, source comparison, and drafting; the human evaluated relevance, tone, cultural grounding, and argumentative coherence; contested claims were checked against primary or reputable institutional sources; and the final judgement remained human. In that sense, this article is a small live demonstration of its thesis: the “hard work” of searching, organising, and synthesising evidence can be accelerated by AI, while the “think work” of framing, judging, contextualising, and deciding what matters remains human.
References
AI Solutions by Davies Meyer. (2026, February 8). Hugging Face. https://ai-solutions.daviesmeyer.com/en/glossary/hugging-face
Bessen, J. (2015, March). Toil and technology. Finance & Development, 52(1). International Monetary Fund. https://www.imf.org/external/pubs/ft/fandd/2015/03/bessen.htm
BharatGen. (n.d.). Param-1: Revolutionizing AI for India — A 2.9B parameter model. https://bharatgen.com/param-revolutionizing-ai-for-india/
Bhaimiya, S. (2023, February 2). ChatGPT may be the fastest-growing consumer app in internet history, reaching 100 million users in just over 2 months, UBS report says. Business Insider. https://tech.yahoo.com/apps/articles/chatgpt-may-fastest-growing-consumer-114524444.html
CFO Brew. (2025, October 30). AI’s generation gap. https://www.cfobrew.com/stories/2025/10/30/ai-s-generation-gap
CXO Digital Pulse. (2025). IndiaAI Mission expands with eight new players to build global AI leaders. https://www.cxodigitalpulse.com/indiaai-mission-expands-with-eight-new-players-to-build-global-ai-leaders/
EPW Research Foundation. (n.d.). Growth of ATMs in India. https://epwrf.in/includefiles/c10816.htm
FM Magazine. (2025, October). Gen Z leads in AI adoption, upskilling, but training gaps persist. https://www.fm-magazine.com/news/2025/oct/gen-z-leads-in-ai-adoption-upskilling-but-training-gaps-persist/
Frey, C. B., & Osborne, M. A. (2013). The future of employment: How susceptible are jobs to computerisation? Oxford Martin School, University of Oxford.
Google DeepMind. (2024, July 25). AI achieves silver-medal standard solving International Mathematical Olympiad problems. https://deepmind.google/blog/ai-solves-imo-problems-at-silver-medal-level/
Google DeepMind. (2025a, May 14). AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms. https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/
Google DeepMind. (2025b, July 21). Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad. https://deepmind.google/blog/advanced-version-of-gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad/
Government of India. (2026a, February 17). Prime Minister Modi has launched BharatGen’s latest models, including the Param-2 text foundation model in 22 scheduled Indian languages [Press release]. Press Information Bureau. https://www.pib.gov.in/PressReleaseDetail.aspx?PRID=2229273
Government of India. (2026b, February 21). Sarvam AI powering a Made-in-India AI revolution [Press release]. Press Information Bureau. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2231169®=3&lang=1
Information Technology and Innovation Foundation. (2022, September 30). Oops: The predicted 47 percent of job loss from AI didn’t happen. https://itif.org/publications/2022/09/30/oops-the-predicted-47-percent-of-job-loss-from-ai-didnt-happen/
LSVP. (2014, May 30). This fourth wave of Indian enterprise software startups is world-class. https://lsvp.com/stories/this-fourth-wave-of-indian-enterprise-software-startups-is-world-class/
Merchant, A., Batzner, S., Schoenholz, S. S., Aykol, M., Cheon, G., & Cubuk, E. D. (2023). Scaling deep learning for materials discovery. Nature, 624, 80–85. https://doi.org/10.1038/s41586-023-06350-6
Ministry of Education. (2020). National Education Policy 2020. Government of India. https://www.education.gov.in/sites/upload_files/mhrd/files/NEP_Final_English_0.pdf
Morgan, J. [@jacobm]. (2026, February 25). The ATM/bank teller story is the most-cited historical analogy in the automation debate [Post]. X. https://x.com/jacobm/status/2026752055063491020
NASSCOM. (2026, January 13). India’s AI talent inflection point: From skill gaps to competitive advantage. https://nasscom.in/knowledge-center/publications/indias-ai-talent-inflection-point-skill-gaps-competitive-advantage
NASSCOM. (n.d.-a). Advancing India’s AI skills: Interventions and programmes needed. https://nasscom.in/knowledge-center/publications/advancing-indias-ai-skills-interventions-and-programmes-needed
NASSCOM. (n.d.-b). India’s services sector and the AI opportunity. https://nasscom.in/voices/indias-services-sector-and-ai-opportunity
National Payments Corporation of India. (n.d.). UPI—Unified Payments Interface. https://www.npci.org.in/what-is-upi
National Women’s History Museum. (n.d.). Women and computing. Retrieved September 26, 2026, from https://www.womenshistory.org/articles/women-and-computing
NITI Aayog, Boston Consulting Group, & NASSCOM. (2025). Roadmap for job creation in the AI economy. https://www.niti.gov.in/sites/default/files/2025-10/Roadmap_for_Job_Creation_in_the_AI_Economy.pdf
Novikov, A., Vũ, N., Eisenberger, M., Dupont, E., Huang, P.-S., Wagner, A. Z., Shirobokov, S., Kozlovskii, B., Ruiz, F. J. R., Mehrabian, A., Kumar, M. P., See, A., Chaudhuri, S., Holland, G., Davies, A., Nowozin, S., Kohli, P., & Balog, M. (2025). AlphaEvolve: A coding agent for scientific and algorithmic discovery (arXiv:2506.13131). Google DeepMind. https://arxiv.org/pdf/2506.13131
OpenAI. (2025, July 19). We achieved gold medal-level performance on the 2025 International Mathematical Olympiad with a general-purpose reasoning LLM [Post]. X. https://x.com/OpenAI/status/1946594928945148246
Reserve Bank of India. (2013). Annual report. https://www.rbi.org.in/scripts/PublicationsView.aspx?id=14932
Sarvam AI. (2025, April 26). Sarvam to build India’s sovereign large language model. https://www.sarvam.ai/blogs/indias-sovereign-llm
Sharma, G. (2025). India’s developmental approach in AI and competences (SWP Research Paper). German Institute for International and Security Affairs. https://www.swp-berlin.org/publications/products/arbeitspapiere/Sharma_WP_August_2025.pdf
Visier. (2022, December 14). To optimize work automation, get beyond “robots taking our jobs.” https://www.visier.com/blog/optimize-work-automation/
Willison, S. (2025, July 19). OpenAI’s gold medal performance on the International Math Olympiad. https://simonwillison.net/2025/Jul/19/openai-gold-medal-math-olympiad/
Wikipedia contributors. (n.d.-a). Automated teller machine. In Wikipedia. Retrieved September 26, 2026, from https://en.wikipedia.org/wiki/ATM
Wikipedia contributors. (n.d.-b). History of computing in India. In Wikipedia. Retrieved September 26, 2026, from https://en.wikipedia.org/wiki/History_of_computing_in_India
Wikipedia contributors. (n.d.-c). Pascaline. In Wikipedia. Retrieved September 26, 2026, from https://en.wikipedia.org/wiki/Pascaline
World Economic Forum. (2025, January 8). Future of Jobs Report 2025: 78 million new job opportunities by 2030 but urgent upskilling needed to prepare workforces [Press release]. https://www.weforum.org/press/2025/01/future-of-jobs-report-2025-78-million-new-job-opportunities-by-2030-but-urgent-upskilling-needed-to-prepare-workforces/
Comments