Dani Rodrik
Artificial intelligence is no longer a distant technological possibility. It is becoming an increasingly influential force in education, employment, public administration, research, business and everyday life. Its ability to process information, generate ideas, solve complex problems and perform tasks traditionally associated with human intelligence is expanding rapidly. What initially appeared to be a sophisticated digital assistant is increasingly capable of undertaking work that once required years of professional training and experience.
This transformation is producing a mixture of excitement, anxiety and uncertainty. Economists, teachers, lawyers, researchers, policymakers and other professionals are confronting a difficult question: if machines can increasingly perform intellectual tasks, what will remain distinctively human about work, creativity and decision-making?
In his recent commentary, The Five Stages of AI Grief, economist Dani Rodrik draws an interesting parallel between society’s response to artificial intelligence and the five stages of grief identified by psychiatrist Elisabeth Kübler-Ross: denial, anger, bargaining, depression and acceptance. His central observation is that people initially underestimated the technology, then became frustrated by its mistakes, subsequently negotiated a division of labour between humans and machines, and eventually confronted the possibility that artificial intelligence could surpass their expectations in areas they had considered uniquely human. The final stage, however, is different from the experience of terminal illness: humanity retains the ability to determine how this technology is developed and used.
That distinction deserves serious attention. Artificial intelligence is not an uncontrollable natural force. It is a human-created technology, developed by institutions, deployed by businesses and incorporated into public systems through political, economic and social decisions. Its consequences will therefore depend not only on what machines can do but also on the rules, incentives and institutions governing their use.
From Technological Curiosity to Intellectual Disruption
The first response to artificial intelligence was often denial. Many assumed that it would remain a more advanced search engine or a chatbot capable of producing convincing but unreliable answers. Its early mistakes appeared to confirm that assessment. Incorrect legal references, fabricated information, unreliable software and superficial academic responses offered reasons to dismiss its capabilities.
But the technology improved. Its errors did not disappear, yet its ability to analyse information, develop arguments, prepare reports and assist with complex professional tasks became increasingly difficult to ignore. The discussion consequently shifted from whether artificial intelligence could perform meaningful intellectual work to how much of that work should be delegated to it.
This transition has exposed an important weakness in conventional assumptions about human intelligence. It was relatively easy to argue that machines could calculate, retrieve information and execute instructions, while humans would retain exclusive control over judgment, imagination, creativity and values. However, as Rodrik illustrates through his own experiments, artificial intelligence can also generate unexpected policy options, suggest alternative ways of framing problems and produce creative material. The boundary between mechanical assistance and intellectual contribution is becoming less obvious.
The challenge, therefore, is not simply that machines are becoming more capable. It is that society must reconsider the meaning of expertise, professional judgment and human contribution in an environment where cognitive work can increasingly be automated or augmented.
The Danger of Outsourcing Human Judgment
The most consequential risk may not be that artificial intelligence replaces every human worker. It may be that people gradually surrender their capacity to think independently because relying on machines becomes easier, faster and economically attractive.
A student who uses artificial intelligence to understand a difficult subject can benefit from personalised explanations and additional learning support. But a student who delegates every assignment to a machine may complete academic requirements without developing the ability to reason, question and construct an argument. Similarly, a lawyer can use artificial intelligence to examine documents and identify relevant legal questions, but professional responsibility cannot be transferred to a system that may produce convincing yet incorrect conclusions.
The same distinction applies to public administration. Governments can use artificial intelligence to analyse data, improve service delivery, identify patterns of public expenditure and evaluate policy options. Yet decisions involving rights, public resources, social priorities and individual freedoms require accountable human judgment. An algorithm may recommend an administrative decision, but citizens must retain the right to understand, challenge and appeal decisions that affect their lives.
Efficiency is not the only measure of good governance. A decision can be fast and technically sophisticated while remaining unjust, discriminatory or inconsistent with constitutional principles. Public institutions must therefore distinguish between using artificial intelligence as an instrument of analysis and allowing it to become an unaccountable authority.
The Political Economy of Artificial Intelligence
Artificial intelligence also raises questions about the distribution of economic power. The benefits of technological progress do not automatically reach society equally. The institutions that own computing infrastructure, control data, develop advanced models and determine access to technological platforms can acquire considerable economic influence.
If productivity increases but the gains remain concentrated among a limited number of corporations and investors, technological advancement may deepen existing inequalities. Workers may face displacement or declining bargaining power, while businesses with access to advanced systems gain advantages over smaller competitors. Countries without domestic technological capacity may become dependent on foreign companies for essential digital infrastructure.
For developing countries such as Pakistan, this issue is particularly important. Artificial intelligence could help improve education, agriculture, healthcare, taxation, urban management and public service delivery. It could also support small businesses, expand access to specialised knowledge and reduce administrative costs. However, these benefits require more than access to software. They depend on reliable digital infrastructure, affordable computing, quality education, skilled professionals, data protection and institutions capable of regulating emerging technologies.
Without these foundations, the country risks becoming a consumer of technologies developed elsewhere, with limited influence over their design, governance and economic benefits. The question is not whether Pakistan should embrace artificial intelligence. It is whether the country can develop the institutional and human capacity to use it productively while protecting its economic and public interests.
Why Regulation and Institutional Responsibility Matter
Rodrik’s most important argument concerns the possibility of choice. Unlike an unavoidable medical condition, artificial intelligence and its applications remain subject to human decisions. Technology companies determine how systems are developed and marketed. Businesses decide which tasks to automate. Professional communities establish standards for responsible use. Governments determine the legal and regulatory environment.
This makes institutional responsibility central to the debate. Regulation should not seek to obstruct innovation or preserve every existing professional arrangement. Nor should it allow technological development to proceed without accountability. The objective must be to establish safeguards that protect human rights, privacy, transparency and public interest while enabling beneficial innovation.
Governments should require appropriate human oversight in high-impact decisions, establish clear responsibility for errors and misuse, protect personal data and ensure that automated systems can be independently assessed. Educational institutions should teach students how to use artificial intelligence without abandoning independent reasoning. Employers should treat technological adoption as an opportunity to improve work and develop skills rather than simply as a mechanism for reducing labour costs.
Most importantly, public policy must address the transition for workers whose occupations may be transformed. Investment in education, professional training and social protection will be necessary to ensure that technological progress does not leave large sections of society economically insecure.
Acceptance Must Not Mean Surrender
The final stage of AI grief should not be interpreted as unconditional acceptance of every technological development. Acceptance should mean recognising the reality of artificial intelligence while retaining the ability to establish limits on its use. It requires societies to abandon both technological denial and blind technological enthusiasm.
Rodrik invokes the story of Odysseus, who ordered his crew to tie him to the mast of his ship so that he could hear the Sirens without being drawn towards their destructive song. The analogy offers a useful lesson: human beings sometimes need institutions and rules to protect themselves from the consequences of their own temptations. Artificial intelligence presents a similar challenge. Its convenience and capabilities may encourage individuals, businesses and governments to delegate more than they should.
The objective must be to ensure that technological power remains subordinate to human purposes. Machines may assist with analysis, generate possibilities and improve productivity, but they should not determine the values by which societies organise themselves. Human dignity, democratic accountability, intellectual independence and public responsibility cannot simply be treated as inefficiencies to be eliminated.
Artificial intelligence will undoubtedly reshape the relationship between people and work. But the future is not determined by technological capability alone. It will be shaped by political choices, institutional safeguards, economic arrangements and the willingness of societies to preserve human agency.
The real question is not whether artificial intelligence will become more powerful. It is whether humanity will become more capable of governing that power.
Dani Rodrik, Professor of International Political Economy at Harvard Kennedy School, is Past President of the International Economic Association and the author of Shared Prosperity in a Fractured World: A New Economics for the Middle Class, the Global Poor, and Our Climate (Princeton University Press, 2025).
