On artificial intelligence, technological displacement, the future of work, universal basic income, and why education must prepare us for more than one version of the future.
By Sahil Tayal | October 2026
“But Sahil, will AI take our jobs?”
It is a reasonable question, and perhaps one of the most important questions people can ask about artificial intelligence today. When a technology becomes capable of writing reports, generating software, analysing information, producing designs and performing tasks that once required specialised professional training, it is natural to wonder what happens to the people who currently earn their living performing those activities.
My answer is that artificial intelligence will almost certainly change the distribution of work. Some tasks will become automated, some occupations will require fewer workers, existing roles will be reorganised, and entirely new forms of work may emerge. The scale, speed and distribution of these changes remain uncertain, and anyone claiming to know exactly which professions will disappear by a particular year is making a prediction that deserves scrutiny.
But I also think the question reveals something about how we have learned to understand education, employment and economic security.
For many people, the immediate objective is understandably straightforward: acquire a qualification, find employment, support a family, meet financial obligations and build a stable life. There is nothing intellectually inferior about this objective. Employment provides income, independence, dignity, routine and access to opportunities that cannot simply be replaced by philosophical reflection.
The person preparing for an interview tomorrow is not asking an abstract question about the future of civilisation. They may be asking whether their education, effort and experience will continue to provide a viable livelihood.
Any serious discussion of artificial intelligence must begin by respecting that reality.
1. Will AI Actually Replace Jobs? What the Evidence Says
The International Monetary Fund estimated in 2024 that almost 40% of global employment was exposed to artificial intelligence, with exposure rising to approximately 60% in advanced economies.
Exposure, however, is not the same as unemployment. Some workers may become more productive because AI assists them. Others may face declining demand for particular skills or tasks, while employers may reorganise teams, reduce hiring or substitute software for certain forms of labour.
The International Labour Organization's 2025 research similarly estimated that approximately one in four workers worldwide held occupations with some degree of exposure to generative AI. Its assessment emphasised that transformation, rather than complete occupational replacement, was the more likely outcome for most exposed jobs.
These findings are neither a guarantee of employment security nor evidence of an inevitable employment catastrophe.
A job is usually a collection of tasks rather than one indivisible activity. An accountant may perform data entry, reconciliation, regulatory interpretation, client communication and professional judgment. A teacher may explain concepts, assess understanding, motivate students, manage classrooms and communicate with families. A software developer may write code, diagnose problems, understand user requirements, make architectural decisions and maintain systems.
AI may perform some of these tasks effectively without automatically replacing the entire occupation.
But the reverse is also possible: if a sufficiently large share of economically valuable tasks becomes automatable, employers may require fewer people to deliver the same output. Productivity gains do not necessarily translate into unchanged employment.
The question is therefore not simply whether AI can perform a task. It is whether automation becomes reliable, economically attractive and organisationally practical enough to change the demand for human labour.
2. We Have Been Here Before, Although Never in Exactly the Same Way
Technological disruption did not begin with ChatGPT.
Agriculture transformed human settlement and the organisation of labour over thousands of years. Writing systems and later printing technologies changed how knowledge could be recorded, preserved and distributed. The printing press, developed in Europe during the fifteenth century, helped expand the circulation of books and ideas while changing the economics of publishing and the authority of institutions that had previously controlled access to written knowledge.
The steam engine and mechanised textile production contributed to industrialisation during the eighteenth and nineteenth centuries. They changed manufacturing, transport, urbanisation and the organisation of working life. Electrification subsequently transformed factories, households, communication and production processes. The telephone altered communication across distance, while computers and the internet changed how organisations processed information, coordinated activity and reached customers.
None of these technologies merely replaced one tool with another.
They helped reorganise occupations, institutions, economic relationships and, over time, cultural expectations.
The historical record also contains genuine displacement and suffering. Mechanisation could undermine established crafts and livelihoods. Industrialisation created new employment opportunities while also exposing workers to dangerous conditions, unstable incomes and profound social adjustment. The benefits of technological progress were neither immediate nor equally distributed.
It would be misleading to conclude from this history that every displaced worker eventually found an equally rewarding occupation. Some did not. Some communities experienced prolonged decline even as aggregate productivity increased.
Equally, it would be misleading to imagine that technological change has only destroyed employment. Entire industries and occupations emerged around electricity, automobiles, telecommunications, computing and digital services.
The lesson is not that history guarantees a happy ending.
It is that technological progress changes the structure of economic opportunity, and the transition itself deserves as much attention as the eventual productivity gains.
Artificial intelligence may differ from earlier waves because it increasingly affects cognitive tasks associated with educated and professional workers. That makes historical analogies useful, but incomplete.
3. “What About the People Who Already Lost Their Jobs?”
This is where technological optimism must confront its most difficult objection.
What about the employee whose role is eliminated? What about the family whose household income depends on that salary? What about someone who spent years developing a specialised skill only to discover that the market no longer rewards it as highly?
Those consequences are real, regardless of whether the technology produces substantial benefits elsewhere.
I do not believe technological change is inherently moral or immoral. A machine does not become ethically good simply because it increases productivity, nor does an innovation become inherently bad because it disrupts an existing occupation. The moral questions arise from how people and institutions develop, deploy and govern technologies, and how they respond to those affected.
A company may have legitimate reasons to automate a process that reduces costs or improves quality. A displaced employee may have equally legitimate reasons to experience that decision as a profound economic loss.
Both realities can exist simultaneously.
Telling a worker that technological progress is inevitable does not pay their rent. Promising that new jobs will eventually appear does not establish that those jobs will be accessible in the same city, require comparable skills or provide similar earnings.
This is why discussions of retraining, unemployment protection, portable benefits, career transitions and accessible education are not peripheral to the AI conversation.
They concern the conditions under which societies adapt to technological change.
The objective should not be to pretend that disruption can always be prevented. It should be to examine how its costs can be reduced and how people can retain meaningful opportunities when the economic value of their existing work changes.
4. What AI Can Do, and What It Cannot Do by Itself
Artificial intelligence is particularly powerful when problems involve information processing, pattern recognition, prediction, classification, generation or the coordination of digital processes.
It can assist with writing, translation, coding, document analysis, research organisation, scheduling, customer support and certain aspects of professional decision-making. In appropriate settings, it can reduce the time required to perform these activities and make sophisticated capabilities available to people who previously lacked the necessary tools or training.
But generating a plan is not equivalent to implementing that plan.
An AI system can help an architect explore building designs, estimate quantities, identify potential conflicts in plans and organise construction schedules. It cannot, as a purely digital system, excavate a foundation, pour concrete, install electrical wiring or inspect a physical structure.
Those tasks require people, machinery, materials, energy, logistics and physical execution.
Of course, robotics changes this boundary. Machines can already perform many physical tasks, and the combination of AI with robotics may automate more of them.
The important distinction is not that physical work will remain permanently beyond automation. It is that software intelligence and physical execution are different engineering challenges, with different costs, risks and constraints.
A program can generate an excellent proposal for expanding clean-water access. That does not automatically secure land, finance infrastructure, maintain pipelines or establish reliable public administration.
An AI tutor can explain mathematics. It cannot guarantee that every learner has electricity, connectivity, nutrition, a safe environment or the support needed to learn.
This is why I distinguish technological capability from institutional capacity. Knowing how to accomplish something does not guarantee that the necessary resources and organisations exist to accomplish it.
5. But Won't Robots Eventually Take Over Physical Work Too?
Some physical automation is already well established.
According to the International Federation of Robotics' World Robotics 2025 report, approximately 542,000 industrial robots were installed worldwide in 2024, with more than 4.6 million industrial robots operating globally that year.
These are not speculative science-fiction machines. Industrial robots already perform tasks such as welding, material handling, assembly, packaging and other forms of repetitive manufacturing work.
Humanoid robotics is also developing, although its current capabilities should not be confused with universal competence in human environments. The International Federation of Robotics' Humanoid Robots: Vision and Reality examines both the ambitions and practical limitations of humanoid systems.
Consider the difference between a robot operating within a carefully designed factory and one navigating an unpredictable construction site. A factory can be organised around repetitive movements, predictable materials and controlled operating conditions. A construction site introduces changing terrain, weather, variable materials, safety hazards, coordination among different trades and unforeseen physical conditions.
Robots may increasingly perform parts of construction, agriculture, logistics, maintenance and domestic work. That does not establish when, or whether, fully general-purpose machines will become economically reliable substitutes for people across all those environments.
Physical automation depends on hardware costs, maintenance, energy, dexterity, perception, safety, regulation and the economics of deployment.
My expectation is that physical automation will continue advancing, but unevenly. Highly structured environments are likely to remain easier to automate than unpredictable settings requiring flexible physical judgment.
That is a forecast, not a timetable.
6. My Forecast: The Future of Work Will Be Uneven
If I were to describe my own expectations, I would distinguish between developments I consider plausible and outcomes that remain deeply uncertain.
I expect the cost of producing many forms of digital work to continue declining. Writing preliminary reports, generating standard software components, preparing routine presentations, processing documents and conducting certain forms of information analysis may require substantially less human effort than before.
I also expect organisations to redesign workflows around these capabilities. In some cases, employees will accomplish more. In others, fewer employees may be needed. New responsibilities may emerge around evaluation, system integration, reliability, security and the management of increasingly automated processes.
Entry-level employment deserves particular attention. If AI performs tasks traditionally assigned to junior employees, organisations may need to reconsider how people acquire the experience required for more advanced work. A profession cannot sustainably depend on experienced specialists while eliminating every pathway through which new specialists develop.
I expect expertise to remain valuable, but its economic rewards may shift. The ability to generate a technically plausible answer may become less distinctive, while the ability to identify the right problem, assess evidence, manage consequences and deliver reliable outcomes may become more important.
However, I would not claim that creativity, empathy or judgment are permanently immune to technological substitution. AI systems may become increasingly capable in activities we currently associate with those qualities. Whether people and institutions choose to delegate particular decisions will depend on performance, cost, trust, regulation and social preferences.
I also do not believe every person must become an entrepreneur, programmer or AI specialist. Societies require many forms of work, and individuals differ in their interests, responsibilities and circumstances.
My forecast is therefore not a single future in which all workers become technologically enhanced knowledge professionals.
It is a future in which technological capabilities advance at different rates, occupations change unevenly, and institutions face difficult decisions about how to distribute the resulting opportunities and risks.
7. Perhaps We Are Asking Education the Wrong Question
This brings me back to an educational question I explored in my earlier essay, Beyond the Classroom: Imagining an Education That Makes Room for Different Minds.
Much of the discussion around AI and education asks which skills students should learn to remain employable ten or twenty years from now.
It is a legitimate question, but perhaps not a sufficient one.
If the future is genuinely uncertain, an educational system cannot confidently prepare every learner for a fixed list of future occupations. It can, however, help people develop the foundational knowledge, adaptability and judgment needed to learn new things throughout their lives.
In the earlier essay, I examined educational approaches including Agora Education, High Tech High, Big Picture Learning and Pratham's Teaching at the Right Level.
These approaches differ substantially in philosophy, evidence and institutional design. They should not be treated as interchangeable or universally superior to conventional schooling.
What interests me is the broader possibility of designing education around both foundational competence and individual development.
A student interested in fragrance might encounter chemistry, sensory biology, psychology and entrepreneurship through that interest. Another fascinated by music might discover mathematics, acoustics, history and technology. A third interested in social behaviour might explore economics, psychology and organisational systems.
Such exploration should complement rather than replace systematic learning.
Literacy, numeracy, scientific reasoning, disciplinary knowledge and practical competence remain indispensable. Without them, students may become dependent on tools whose outputs they cannot evaluate.
UNESCO's Guidance for Generative AI in Education and Research similarly advocates a human-centred approach to educational technology.
My argument is that education should prepare people not only to qualify for their first occupation, but also to develop the capacity to navigate changes that nobody can forecast with certainty.
That includes technical competence, but also critical thinking, emotional literacy, cooperation, ethical judgment and the ability to distinguish persuasive information from reliable knowledge.
8. But Not Everyone Has the Luxury of Planning Twenty Years Ahead
This is an important qualification.
It is easy to discuss intellectual exploration, lifelong learning and the future of civilisation from the perspective of someone who has sufficient time, resources or security to engage in those questions.
For many people, education is closely connected to immediate economic mobility. A student may be preparing for competitive examinations because employment would materially improve their family's circumstances. A working professional may be studying for a certification because it offers a credible path to a promotion. A parent may prioritise stable income over experimentation because other people depend on that income.
These are not failures of imagination.
They are reasonable responses to real responsibilities and constraints.
I would not want an educational philosophy that celebrates unconventional learning while dismissing the importance of employability. Nor would I want to romanticise uncertainty for people who bear greater financial risks than I do.
The objective should be to expand people's choices, not to prescribe a supposedly enlightened lifestyle.
A strong educational system should support those seeking immediate employment, those developing specialised expertise, those pursuing academic inquiry and those exploring unconventional interests.
It should also recognise that individuals may move between these objectives at different stages of life.
The future of education should not require everyone to become the same kind of person.
9. Universal Basic Income: What Happens If Productivity Rises but Labour Demand Falls?
One proposal that becomes relevant in this discussion is universal basic income, or UBI.
Under a conventional UBI model, individuals receive a regular cash payment without employment requirements or means testing. The World Bank's guide to universal basic income examines its conceptual foundations, potential benefits and practical challenges.
The attraction is understandable. If technology enables societies to produce more goods and services with less human labour, people may ask whether economic security should remain as tightly connected to conventional employment as it is today.
A basic income might provide a financial floor, reduce certain forms of insecurity and give people greater flexibility during occupational transitions.
But the proposal also raises difficult questions.
How would it be financed? What level of payment would be meaningful? Would universal transfers be more effective than targeted support? How would they interact with healthcare, pensions, disability benefits, education and existing welfare programmes? What effects might different designs have on employment, prices and public finances?
Pilot programmes and cash-transfer research can help investigate particular outcomes, but they do not establish that a permanent, nationwide UBI would produce identical results in every economy.
It is also important to distinguish UBI from unemployment insurance, guaranteed minimum income, wage subsidies and targeted cash transfers. These policies may share certain objectives while operating through substantially different mechanisms.
I do not regard UBI as an inevitable consequence of artificial intelligence, nor do I think it can be dismissed without examination.
If AI substantially changes the relationship between productivity and employment, societies may need to reconsider how income security, economic participation and public services are organised.
UBI is one possible response among several, not a complete solution to the future of work.
10. Before Imagining an AI-Powered Civilisation, Consider the World We Already Have
There is a final dimension to this discussion that deserves humility.
According to the WHO and UNICEF's 2025 report, approximately 2.1 billion people lacked safely managed drinking water in 2024.
This does not mean all 2.1 billion people had no access to water. Safely managed drinking water is a specific international development indicator involving an improved source that is accessible on premises, available when needed and free from contamination.
The same data also document progress: approximately 961 million people gained access to safely managed drinking water between 2015 and 2024.
Both facts matter.
We are developing increasingly capable AI systems while a substantial share of humanity still lacks access to basic infrastructure that developed societies may take for granted.
This is not an argument against artificial intelligence, robotics or advanced scientific research.
It is a reminder that technological possibility and human welfare are not identical.
An AI system may help engineers optimise water networks, identify leaks, forecast demand or improve infrastructure planning. Yet providing safe water still requires financing, construction, maintenance, governance and effective institutions.
The same distinction applies to education, healthcare, sanitation and housing.
A society may possess extraordinary computational intelligence while struggling to coordinate the delivery of solutions it already understands.
That is why I am cautious about claims that increasingly powerful intelligence will automatically solve humanity's most persistent problems.
Intelligence can help us identify possibilities. Turning those possibilities into widely shared benefits requires additional forms of capability.
11. What Should We Actually Prepare For?
I cannot tell a student which occupation will be completely secure in 2040. I cannot promise that every worker displaced by AI will find an equally rewarding job, and I cannot establish that universal basic income will become economically or politically viable.
What I can suggest is a different way of approaching uncertainty.
Rather than assuming that education ends when employment begins, we can recognise learning as a continuing part of adult life. Rather than treating every new technology as either salvation or catastrophe, we can examine its actual capabilities, limitations, costs and consequences. Rather than measuring human development only through productivity, we can also consider security, autonomy, health, relationships and meaningful participation in society.
These are not substitutes for practical economic planning. They are broader considerations that can help inform it.
I am optimistic about artificial intelligence because it can reduce barriers to learning, creation and problem-solving. I am cautious about its employment effects because technological gains can be distributed unevenly. And I remain uncertain about the ultimate structure of future labour markets because the relevant technologies, institutions and economic decisions are still evolving.
Perhaps the most responsible answer to the original question is therefore neither reassurance nor alarm.
Yes, AI will take over some work that people currently perform. It may eliminate some jobs and create others. But the future of employment will depend not only on what machines become capable of doing, but also on how human societies decide to organise work, opportunity and economic security around those capabilities.
And perhaps education's greatest contribution will not be teaching every child to predict the future.
It will be helping more people develop the knowledge, confidence, adaptability and judgment to participate meaningfully in whatever future arrives.
Sources, References and Further Reading
Artificial intelligence and employment
International Monetary Fund (2024). AI Will Transform the Global Economy. Let's Make Sure It Benefits Humanity.
International Labour Organization (2025). Generative AI and Jobs: A 2025 Update.
Robotics and physical automation
International Federation of Robotics (2025). World Robotics 2025: Global Robot Demand in Factories Doubles Over Ten Years.
International Federation of Robotics (2025). Humanoid Robots: Vision and Reality.
Education and human development
UNESCO (2023). Guidance for Generative AI in Education and Research.
Pratham — foundational learning and Teaching at the Right Level.
High Tech High — project-based learning.
Big Picture Learning — personalised and interest-driven education.
Agora Education — student-centred educational approaches.
Economic security and basic needs
World Bank (2020). Exploring Universal Basic Income: A Guide to Navigating Concepts, Evidence, and Practices.
World Health Organization and UNICEF (2025). One in Four People Globally Still Lack Access to Safe Drinking Water.
UNICEF Data. Access to Drinking Water.
Related independent writing
Author's note: This is an independent interdisciplinary essay, not a labour-market forecast produced through econometric modelling. Institutional statistics and research findings are attributed to their sources. Predictions about future employment, education and robotics are my own interpretations and remain open to revision as stronger evidence becomes available.
Preserved from the LinkedIn data export of 24 October 2026. Links and references reflect the original article; they are not independently re-verified here.