Honey, we upskilled the kids! But did we prepare them for the future?

“But, what about the children?” is a common, and in many instances, understandable, guiding question for how we approach phenomena that have the potential to disrupt – from something as individual as divorce in the family to something as societal as laws around poverty. AI is no different. The disruption resulting from the development and deployment of generative AI across every aspect of our lives – school, work, leisure, relationships, the arts – compels us to face the present we are building and the future we are laying the foundation for. Especially, for the children.In India, the prevailing answer is: invest in AI upskilling. This was a dominant theme at the India AI Impact Summit 2026, and its logic pervades both formal and informal learning. AI upskilling initiatives abound: revamping curricula for grades 3-8 to include courses on computational thinking and AI,(1) equipping teachers with the skills to integrate AI into classroom preparation and teaching,(2) and offering professional courses for informal workers in rural India. The message is clear, there is no room for interpretation: to be AI-ready is to be future-ready.(3)But this assumes we know what future we are preparing young people for. Do we, though?

A market that is moving faster than the preparation

The optimistic version of the answer is straightforward: the future is AI, and therefore, AI proficiency will be rewarded. In white-collar fields such as IT services, the expectation is that human workers will collaborate closely with AI, through productivity tools or agents, to accomplish tasks and realize the benefits. AI upskilling can potentially make huge inroads here – if you can secure the job in the first place!Young adults globally are entering a job market undergoing significant restructuring. The pressure to adopt AI has resulted in massive layoffs at established companies(4) and a shrinking number of entry-level jobs for an increasingly college-educated workforce.(5) The irony is sharpest for young adults in India who strategically pursued STEM-related fields, such as software engineering, as a direct path to upward mobility. Indian IT service companies have reduced hiring for entry-level roles by 20-25% due to automation,(6) while Big Tech companies have reduced hiring by 50% over the last three years.(7) Roles with automatable workflows(8) are likely to follow a similar trajectory, deploying agentic AI (autonomous AI systems capable of executing tasks without constant human oversight) to talk with customers, generate leads, and schedule doctor appointments.(9)And even for those who do get in, the nature of the work has shifted in ways that carry their own costs. Where entry-level engineers once had to write their own code, they are now expected to be able to “refactor”, or correct, AI-generated code. This is not a trivial distinction. Building something from scratch is also the process of building expertise. Relegating workers, especially entry-level workers, to reviewing and reworking AI rather than performing the tasks themselves limits the knowledge development that underpins career growth. The coast isn’t clear for those who may have seemed to have secured a job that goes beyond the menial task of overseeing AI: the increasing pressure to be constantly productive, to avoid the AI-driven chopping block, is literally killing them.Meanwhile, AI has generated an entirely different category of work: training itself. Content moderation and data annotation have emerged as relatively accessible sources of income for young people from rural and marginalized communities with limited alternatives. The downside of such work, if one were to take it up, is significant: lack of skill transferability;(10) the shift to higher-skilled data annotation disadvantaging less-skilled data annotators;(11) meagre and irregular pay; and frequent, nonstop exposure to harmful content,(12) with minimal mental health support. As AI increasingly takes over data annotation tasks, even this precarious foothold may shrink.(13) The question of whether AI development is creating more jobs for young adults is less important than asking: jobs of what kind, and at what cost?

The skills paradox

Young people awaiting their turn in the workforce, priming themselves to be the most suitable candidates by focusing on STEM-related subjects or AI upskilling, are still at risk of being systematically disadvantaged by notions of AI’s return on investment. On the one hand, they’re told to keep pace with AI and demonstrate proficiency. On the other hand, they’re encouraged to cultivate the distinctly human skills that machines supposedly cannot replicate: higher-order critical thinking, creativity, socio-emotional intelligence, and complex problem-solving. The implicit promise is that these will serve as a durable differentiator in an AI-saturated labour market.The evidence, however, is not reassuring on either front. Studies on the impact of AI for problem-solving on human cognition have shown signs of cognitive decline in users; separately, research suggests that people often perceive AI responses as more empathetic than human ones. If AI is eroding the very capacities we’re counting on humans to develop – and if people are beginning to prefer AI’s emotional register to our own – neither STEM proficiency nor the “human skills” proposition offers the clear competitive edge it promises.This leaves “AI-native” young adults entering the workforce in a genuine predicament: they could capitulate to market demands and upskill, as many have done by prioritising STEM education over the humanities or social sciences, but the market may shift in ways that render that investment obsolete before they’ve seen returns on it. And with industries and technologies changing rapidly, upskilling at the beginning of the year could mean very little, even a few months later. If they do lean heavily into AI tools, they risk becoming dependent on them in ways that blunt independent judgment. And if they anchor their differentiation in human skills, those very skills may atrophy from disuse – or simply fail to be valued. There is no clean path forward. That is precisely why the framing of “AI readiness” as a preparation problem that better curricula can solve is insufficient on its own.What’s missing is not more preparation – it’s a different kind of thinking altogether.

Learning to see what’s coming

What we need, alongside technical literacy, is a capacity to think seriously about the future – not to predict it, but to interrogate it. This is not a supplementary skill. In a landscape where the rules of work, knowledge, and opportunity are being rewritten faster than curricula can respond, the ability to read emerging change and reason about its consequences may be the most critical thing we can teach.This is where the practice of foresight becomes relevant. Foresight begins with signals: small, concrete innovations, disruptions, or events that indicate emergent change in behaviours, technologies, or policies, and that point toward possible future trajectories. Individually, a signal might seem minor. Triangulated across sectors, geographies, and timelines, signals begin to reveal the shape of change already underway.This is the method that underpins Human in the Loop, a participatory foresight and storytelling project by Digital Futures Lab that explores the near-future risks of generative AI in India. Across six fictional stories spanning agriculture, healthcare, the judiciary, the environment, relationships, and cultural homogenization, Human in the Loop invites everyday readers to sit with the uncomfortable reality of what may become of us and the world we live in, if we allow AI to develop at its present pace and in the present direction.

The scenarios are not speculative for their own sake. To name a few: a frontline health worker experiences massive encroachment on their personal time with patients. A stratified justice system, where more affluent defendants can use generative AI to create heartfelt testimonies (which has happened in real life!).(14) An overly-optimised food production system, where everything tastes uniformly of nothing. An AI companion mediating every feeling and interaction with your closest friend. Each of these stories is built from real signals visible today; the discomfort they produce is the point. Failing to anticipate future consequences early on risks locking us into technological and policy trajectories that become very difficult to reverse. While some disruptions are more obvious, especially in the aggregate, many long-term impacts will emerge more gradually, and perhaps, less obviously.

What foresight can do in a classroom

For teachers, this is not an abstract concern. The children in classrooms today are the ones who will inherit – and be shaped by – the futures these signals are pointing toward. And teachers are unusually well-positioned to cultivate the kind of anticipatory thinking that goes beyond “learn to code” or “develop soft skills”.Foresight methods offer a structured, generative way to do this. A few approaches that translate readily into classroom practice:

• Signal scanning asks students to identify and interpret early signs of change in their immediate environment – a news story, a shift in how a family member’s work has changed, an app that didn’t exist two years ago. When discussed collectively, these become the raw material for thinking about cause, consequence, and trajectory.
• Scenario building invites students to take a current trend or technology and extrapolate it into multiple possible futures – not just the optimistic or the dystopian, but the ambiguous and uneven ones that are most likely. This builds the habit of holding complexity without collapsing it into simple narratives.
• Speculative writing and storytelling, in the tradition of Human in the Loop, asks students to imagine a specific character navigating a specific future shaped by AI – and in doing so, to think concretely about who benefits, who is harmed, and what choices led there.

These exercises are not a detour from the curriculum – they are a corrective to its current blind spots. AI upskilling and computer science education, as they are largely being designed today, treat the future as a fixed destination that students need to be equipped to arrive at. Foresight treats the future as something still being made – and insists that young people have a role in making it. That is a meaningful difference, and it has real implications for how we think about what computer science education is actually for.

Teaching children to code, to work with AI tools, to think computationally – none of this is wrong. But if it is offered without the critical and imaginative scaffolding to question what these technologies are doing to work, to knowledge, to human relationships and to the distribution of opportunity, then we are producing capable operators of a system they have no framework to interrogate or challenge. The signals we see today – shrinking entry-level jobs, the cognitive costs of AI dependence, the precarity of AI-generated work for the most marginalised – are not aberrations to be smoothed over with better upskilling. They are the system working as designed. Children deserve an education that helps them see that.

For teachers navigating the demands of a computer science syllabus while also watching their students grapple with enormous uncertainty about their futures, foresight methods offer something rare: a way to make the uncertainty itself the subject of serious, structured inquiry. The goal is not to frighten students or to cultivate cynicism, but to produce something more durable than optimism: a will to question, and to want more and better for their future. The task now is to build a generation of young people who can look at the world AI is building, ask hard questions about who it serves and still believe that a different future is worth imagining – and worth fighting for.This article was originally published online for Teacher Plus Magazine. You can access the original article here. References

1. “CBSE Mandates AI and Computational Thinking Training for 2026-27, New Curriculum Rolled Out for Classes 3-8.” Times of India, April 2026. https://timesofindia.indiatimes.com/education/news/cbse-mandates-ai-and-computational-thinking-training-for-202627-new-curriculum-rolled-out-for-classes-38/articleshow/130172568.cms
2. “Upskilling India for the AI Transformation.” Observer Research Foundation (ORF), December 15, 2025. https://www.orfonline.org/expert-speak/upskilling-india-for-the-ai-transformation
3. “Empowering India’s Future: How Government AI Upskilling Is Revolutionising Jobs and Governance.” Wadhwani Foundation, December 29, 2025. https://wadhwanifoundation.org/how-government-ai-upskilling-is-revolutionizing-jobs-and-governance/
4. “Some Big-Name Companies Are Laying Off Workers. Here’s What It Means.” ABC News, February 4, 2026. https://abcnews.go.com/Business/big-companies-laying-off-workers-means/story?id=129630523
5. “What Is College for in the Age of AI?” New York Magazine / Intelligencer, January 20, 2026. https://nymag.com/intelligencer/article/what-is-college-for-in-the-age-of-ai.html
6. “Entry-Level Tech Demand Falls 20-25% as India Shifts to a Diamond-Shaped Workforce: EY.” Moneycontrol, November 17, 2025. https://www.moneycontrol.com/artificial-intelligence/entry-level-tech-demand-falls-20-25-as-india-shifts-to-a-diamond-shaped-workforce-ey-article-13679052.html
7. “The SignalFire State of Talent Report – 2025.” SignalFire, May 20, 2025. https://www.signalfire.com/blog/signalfire-state-of-talent-report-2025
8. “AI Leads to Job Cuts in Customer Support, But Will Push for Agentless Setup Hold?” Entrepreneur India, November 4, 2025. https://www.entrepreneur.com/en-in/news-and-trends/ai-leads-to-job-cuts-in-customer-support-but-will-push-for/499163
9. “The Agentic Leap: How India Is Rewiring Business with Autonomous AI.” Salesforce India, February 23, 2026. https://www.salesforce.com/in/news/stories/powering-agentic-era-for-enterprises-in-india/
10. Siddiqui, Zuha. “They Wanted Careers in Tech. They’re Stuck as TikTok Moderators.” Rest of World, February 15, 2024. https://restofworld.org/2024/content-moderators-jobs-pakistan/
11. “Side Hustle or Scam? What to Know About Data Annotation Work.” TIME, April 2, 2024. https://time.com/6962608/data-annotation-legit-tech-jobs-ai/
12. “‘In the End, You Feel Blank’: India’s Female Workers Watching Hours of Abusive Content to Train AI.” The Guardian, February 5, 2026. https://www.theguardian.com/global-development/2026/feb/05/in-the-end-you-feel-blank-indias-female-workers-watching-hours-of-abusive-content-to-train-ai
13. “Is Data Annotation Dying?” Analytics India Magazine, December 24, 2025. https://analyticsindiamag.com/ai-features/is-data-annotation-dying
14. “AI-generated Evidence is a Threat to Public Trust in the Courts.” The National Center for State Courts, February 24, 2026. https://www.ncsc.org/resources-courts/ai-generated-evidence-threat-public-trust-courts#%3A~%3Atext%3DWhen%20California%20Judge%20Victoria%20Kolakowski%20reviewed%20a%2Cv.%20Cushman%20%26%20Wakefield%2C%20something%20felt%20wrong