Opinion Piece: Unlocking the Best and Worst Anxieties of the AI Era – How Taiwan Can Seize the Next Economic Long Wave Through Institutional Innovation

"It was the best of times, it was the worst of times; it was the age of wisdom, it was the age of foolishness." More than 160 years ago, Dickens used these words to describe the contradictions and divisions of a revolutionary era. Today, standing at a historical turning point of rapidly spreading generative AI, humanity once again faces collective anxiety over progress and disorder coexisting. On July 13, 2026, over two hundred economists and AI researchers, including sixteen Nobel laureates, jointly issued a statement urging nations to prepare policies and institutions for the economic transformation AI may bring. The statement warns that AI could trigger economic transformation on a larger scale than the Industrial Revolution, but in a much shorter timeframe. Co-initiator Anton Korinek, a professor at the University of Virginia, further cautioned that past major technological revolutions often gave society decades to adjust—AI may leave only a few years. If action is delayed until all impacts are clear, institutional adaptation may come too late. Why do these experts at the forefront of theory and technology issue such urgent calls? Faced with a force capable of drastically compressing search, analysis, and content generation time in certain knowledge tasks, people often blindly embrace the 'best of times' or fall into the anxiety of being replaced—the 'worst of times.' However, remaining trapped in a binary view of good versus bad easily leads one to lose direction in the flood of technology. By re-examining this transformation through four layers—phenomenon, conditions, ecosystem, and institutions—we discover that AI is not merely a technological revolution, but potentially a key driver of the next economic long wave. The true global competition is extending beyond computing power and model development into governance and institutional capability. First Layer of Thinking: Phenomenal Duality and Moving Beyond the Anxiety of Replacement in the Fifth Wave At the most intuitive level of phenomena, we observe extreme opposition and disruption. From design and programming to customer service, many highly trained tasks are being restructured, accelerated, or partially automated. Jobs may not disappear immediately, but job boundaries, skill value, and organizational division of labor are already shifting. When facing unknown technologies, people often rely on past experiences to understand the future, simplifying AI into a binary choice of 'replacing humans' or 'saving humans.' Many enterprises continue the management inertia formed during the fifth wave of information and communication technology, primarily measuring technological value through cost reduction, efficiency gains, process automation, and labor substitution. Continuing to treat this new generation of technology with a single logic focused on process efficiency easily traps us in the anxiety of gain and loss, and may further widen income and opportunity gaps between those who control computing power, data, and platforms, and other groups. Second Layer of Thinking: Recognizing Conditions—The Deeper Risk Lies in Institutional Lag Experts' deeper concern lies in the Collingridge Dilemma in technology governance: when AI has not yet been widely adopted, governments struggle to grasp its full impact; by the time unemployment, power concentration, or liability disputes become apparent, the technology and business models may already be deeply embedded in society, drastically increasing reform costs. The social consequences of technology are shaped collectively by data sources, commercial incentives, deployment methods, organizational capabilities, and public institutions—not solely by the model. The ultimate impact of AI depends not only on what the model can do, but also on how businesses, governments, and society choose to deploy it. Today, model capabilities continue to accelerate, while institutional adjustments require legislation, negotiation, and social trust. This time gap is creating new governance risks. This appeal reminds us that beyond the risks of the models themselves, the more difficult challenge is that institutional preparation cannot keep pace with technological diffusion. When responsibility, rights, and distribution mechanisms remain stuck in the old era, AI's benefits struggle to become widespread, while its impacts may first be borne by workers, consumers, and vulnerable groups. If existing labor, healthcare, and financial regulations are used to address new types of problems brought by AI, on one hand, valuable applications may struggle to scale; on the other, it may easily amplify controversies over unclear responsibility, risk externalization, and unequal benefit distribution. Third Layer of Thinking: Ecosystem Integration—From Single-point Automation to a Super Ecosystem Drawing on the Kondratiev wave of economic long waves, the fifth wave of ICT revolution primarily advanced information connectivity and process efficiency; we may now be entering a new wave of technological and institutional change driven jointly by AI, biotechnology, energy transition, and intelligent systems. The next phase of competition will further test whether societies can integrate AI, biotech, and green energy into sustainable institutional and ecosystem frameworks. As AI gradually takes on certain computational, analytical, and generative tasks, we can direct its capabilities toward complex problems previously difficult to address due to fragmented data, collaboration challenges, or high costs, forming sustainable and governable systemic solutions. For example, integrating genetic and lifestyle data in healthcare to repay the 'health debt'; optimizing highly complex green energy grids in environmental fields to repay the 'environmental debt.' By constructing cross-domain 'Super Ecosystems,' AI will create not only labor substitution and process efficiency, but also new service models, collaboration relationships, and economic value. Fourth Layer of Thinking: Institutional Restructuring—Transforming Judgment into Governance Capacity After recognizing the issues, the next step is transforming judgment into institutional capacity. For Taiwan, the real challenge lies in whether it can transform its technological advantage in producing AI hardware into institutional arrangements that are clear in responsibility, shared in benefits, and scalable. Taiwan possesses globally leading semiconductor manufacturing, server supply chains, and engineering capabilities, yet still faces another set of questions: When an AI-assisted diagnosis results in error, how should responsibility be shared among model developers, medical institutions, and clinical staff? When companies use employee data, work records, or professional outputs to train models, how should notification, authorization, compensation, and opt-out rights be defined? When small and medium enterprises cannot establish complete governance teams, can the government provide shared testing, validation, and risk management infrastructure? These questions determine whether Taiwan can evolve from an AI hardware supplier into an exporter of AI institutions and application models. It is commendable that Taiwan is not starting from scratch: the 'Artificial Intelligence Basic Act' has been promulgated and implemented, and the National AI Strategy Special Committee, risk classification frameworks, and cross-ministerial regulatory adjustments have been launched. The real test in the next phase is whether governance principles can be transformed into industry-specific, executable, verifiable, and accountable institutions. Governments and enterprises should deepen strategic alignment on existing foundations and advance the following four institutional innovations: Talent Transformation and Social Safety Nets: Establish skill transition accounts and on-the-job training systems; consider wage insurance, reemployment allowances, or transitional income support to avoid measuring AI's social impact solely through unemployment rates. Data, Intellectual Property, and Liability Systems: Properly handle training data licensing, professional knowledge compensation, and algorithmic liability, and establish corresponding insurance mechanisms for high-risk applications such as medical AI and autonomous driving. Cross-domain Regulatory Sandboxes and Demonstration Fields: Allow healthcare, finance, transportation, and government services to conduct trials under clear conditions, using empirical results to feedback and revise existing regulations. Government Procurement and Public Demand Steering: Beyond playing the role of regulator, governments should create the first markets and demand for trustworthy AI through public procurement, standard setting, and data infrastructure development. Conclusion: From Computing Power Advantage to Institutional Advantage Technology can rapidly enter society, but questions of how its value is distributed and who bears the risks must be answered by institutions. As model capabilities and computing power services become increasingly platformized, the duration of maintaining a single technological advantage may shorten. Historical experience shows that whether technological advantage can be transformed into long-term competitiveness often depends on whether institutions can keep up in time. Nations that enable AI to augment human capabilities, fairly distribute value, and clearly assume responsibility will be better positioned to lead the next wave of industrial transformation. This is a competition moving from computing power to governance. Taiwan already stands at a critical position in the global AI hardware supply chain. The next step is to prove that it also has the capability to become an exporter of institutions, standards, and application models. *The author is a researcher in public value creation, innovation ecosystems, and AI governance.