The Work That Arrived by Morning

One of the most memorable scenes in Thomas Friedman’s The World Is Flat involves an American executive completing his day while a remote assistant in India begins another. Research, calculations, or a PowerPoint presentation can be sent across the world in the evening and returned by the following morning. The work continues while the executive sleeps. Distance, once an obstacle to business, has been converted into an advantage.

When Friedman published the book in 2005, this seemed to capture something new about the global economy. The internet had existed for years, but faster networks, fiber-optic connections, personal computers, and compatible business software were allowing companies to divide office work across countries on a practical scale. A document could travel without its author. A customer in North America could speak to a service representative in Manila. An engineer in Bangalore could support a company whose headquarters stood thousands of kilometers away.

The appeal extended beyond efficiency. For countries such as India and the Philippines, globalization appeared to offer a path toward economic development that did not depend entirely on manufacturing or the export of natural resources. Educated workers could participate in the international economy while remaining in their own countries. English proficiency, technical education, and lower living costs became national economic assets. Work that had once been geographically protected in the United States, Europe, or Japan could now be performed wherever qualified people and reliable networks were available.

Two decades later, the world remains flat, but the overnight PowerPoint story has acquired a different meaning. An executive can now ask an AI system to produce a first draft within minutes. A human team may still need to verify the information, improve the argument, correct the design, and assume responsibility for the finished material. Yet the advantage of having another person work through the night is no longer as decisive. The task does not always need to cross the ocean because part of it may no longer need to reach another human being.

From Empires to Individuals

Friedman organized the history of globalization into three broad periods. His categories were never intended as a complete economic history. They offered a way to identify the main actor in each stage and to describe how technology reduced the practical significance of distance.

Globalization 1.0 extended from the age of European exploration to the early nineteenth century. Countries and empires were the principal actors. They crossed oceans in pursuit of territory, trade, resources, religious expansion, and political power. Ships, navigation, military force, and state organization allowed nations to project themselves beyond their borders. In Friedman’s language, the world shrank from large to medium.

Globalization 2.0 began with industrialization and continued into the late twentieth century. Companies became the central agents. Steamships, railways, telegraph systems, telephones, automobiles, aviation, and eventually computers allowed businesses to acquire materials, manufacture products, and sell them across several countries. Production could be separated into stages and distributed through international supply chains. The world moved from medium to small because companies could organize economic activity across distances that had once made such coordination impossible.

Globalization 3.0 emerged around the beginning of the twenty-first century. Individuals and small groups gained tools that had previously belonged mainly to governments and large corporations. A person with a computer, an internet connection, and suitable skills could participate in global work from almost anywhere. At an IMF forum in 2005, Friedman described this as the transition from countries globalizing, to companies globalizing, and then to individuals and small groups entering the global field directly.

Outsourcing became one of the characteristic practices of this third period, although several related terms are often used interchangeably. Outsourcing refers to assigning work to another company. Offshoring means moving work to another country, even when it remains within the same corporate organization. Business process management is broader. It concerns the design, operation, measurement, and improvement of a process, regardless of whether the work is outsourced. The later preference for IT-BPM over BPO reflected an ambition to move beyond call centers and routine back-office work toward managing more complex operations.

The economic imagination of the period was also visible in the language of emerging markets. Goldman Sachs introduced BRIC in 2001 to describe Brazil, Russia, India, and China as economies expected to carry greater weight in the world. South Africa joined the political grouping later, turning BRIC into BRICS. The Next Eleven followed in 2005, including the Philippines, Indonesia, Vietnam, Bangladesh, South Korea, Mexico, and several other countries. These categories were investment concepts rather than coherent development models, but they expressed the confidence of the time. Countries outside the traditional centers of wealth appeared ready to claim a larger share of global growth.

When Work Crossed the Border Instead of the Worker

The Philippine experience gave this transformation a distinctive social meaning. For generations, participation in the international economy often required Filipinos to leave the country. Nurses, engineers, seafarers, construction workers, domestic helpers, and other professionals travelled to North America, Europe, the Middle East, and East Asia. Their remittances supported households, education, property purchases, and national consumption, but the economic arrangement also separated families for years at a time.

The rise of BPO reversed the direction of movement. Instead of sending a Filipino worker to a foreign workplace, a company could send foreign work to the Philippines. The worker stayed in Manila, Cebu, Clark, or another Philippine city while serving customers and organizations abroad. English proficiency remained an advantage, but it could now create employment at home.

This was not a small alteration to the labor market. Call centers, technical-support operations, accounting services, software development, medical transcription, content moderation, and administrative processing expanded across the country. Makati and Ortigas grew as established business districts, while Eastwood and BGC became symbols of a newer service economy. Offices remained active throughout the night because Philippine workers were following business hours in North America and Europe. Around those offices grew condominiums, restaurants, transportation services, retail businesses, and a culture shaped by unusual working schedules.

I saw the same economic logic in the technology industry. Trend Micro developed malware-analysis capabilities in Manila, where Filipino engineers could contribute to an international security operation. This was not conventional outsourcing when the work remained inside the company, but the principle of offshoring still applied. Technical work did not have to remain near the corporate headquarters. A global network could distribute analysis across locations, combine different time zones, and draw on skilled employees whose cost was lower than it would have been in Japan or North America.

India followed a related path with a greater concentration of software engineering, information technology, consulting, and large outsourcing providers. China’s path was different. Its rise depended much more heavily on manufacturing, infrastructure, foreign investment, industrial policy, and integration into global supply chains. Factories producing goods for international brands contributed to its reputation as the factory of the world. Over time, however, China accumulated engineering knowledge, capital, supplier networks, and technological capacity. It did not remain a location where other countries placed less expensive work.

The gains created by this system were genuine, but they were never evenly distributed. An office job created in Manila or Bangalore could correspond to a position eliminated or never created in a more expensive city. Workers in developed economies discovered that education and white-collar status no longer protected them from international competition. Global labor arbitrage, the difference in wages paid for comparable work in different countries, became a central business strategy.

From the perspective of the Global South, the same process offered income, mobility, and access to international careers. From the perspective of displaced workers in developed countries, it could feel like the erosion of security. The flat world expanded the field of opportunity while exposing more people to competition. Its benefits and injuries were two expressions of the same structure.

When the Task Stops Travelling

Artificial intelligence changes that structure because it introduces a third possibility. A task no longer has to remain with the expensive worker or move to the less expensive worker. Parts of it can be transferred to a machine.

Earlier automation had already transformed factories and administrative work. Rule-based software processed transactions, robotic process automation moved data between systems, and automated menus handled basic customer requests. Generative AI extends automation into language, images, analysis, programming, and other activities associated with educated office workers. It does not need every situation to be reduced to an explicit set of rules. It can respond to instructions, interpret unstructured material, generate alternatives, and work across several forms of information.

The old PowerPoint example shows the difference. Under Globalization 3.0, an executive sent notes to India, where another person researched the topic and built the presentation. In the current environment, the executive can give those notes to an AI model. The system may produce an outline, draft the text, recommend charts, create images, and suggest a visual structure before the executive has finished a cup of coffee.

Human participation does not disappear. The AI may misunderstand the audience, fabricate a source, weaken a nuanced argument, or produce a polished presentation with little substance. Someone must decide what should be said, determine whether the claims are reliable, recognize cultural and organizational sensitivities, and accept responsibility for the outcome. Even so, one person working with AI may complete a task that once required several people or many hours of distributed labor.

This makes routine and standardized BPO work especially vulnerable. Basic customer inquiries, data entry, transcription, document processing, scripted technical support, preliminary research, simple translation, and rule-based quality checks are close to the capabilities that AI is improving fastest. The economic advantage of employing a lower-cost worker remains relevant only when that worker contributes enough judgment, context, trust, or specialized knowledge to justify human participation.

The available data does not show that Philippine IT-BPM has already entered an industry-wide collapse. Employment reached about 1.9 million workers in 2025, with revenue of roughly $40 billion. Yet the expected path has changed. In July 2026, the industry association reduced its 2028 employment projection from around 2.5 million to a range between 1.85 million and 2.14 million. Its revenue projections were also revised downward, while the desired direction shifted toward higher value per employee.

These figures describe a more complicated development than mass replacement. Revenue may continue to rise while employment grows more slowly. Companies can reduce their workforces through attrition, hiring freezes, internal transfers, and the elimination of positions that would otherwise have been created. A sector can appear healthy in aggregate statistics while individual workers face insecurity and younger applicants find fewer points of entry.

The ILO’s 2026 study of the Philippines estimated that 27.2 percent of employment had some exposure to generative AI, while 3.6 percent fell within its highest-exposure category. An IMF analysis similarly distinguished between jobs likely to be complemented by AI and those more susceptible to displacement. Exposure does not predict a layoff, but it identifies where tasks and staffing models may change.

One of the least visible consequences may be the weakening of professional apprenticeship. Junior employees have traditionally learned through routine assignments. New analysts classify common cases before handling unusual ones. Young translators develop judgment by working through ordinary documents. Entry-level programmers fix limited problems before designing complex systems. If AI absorbs the routine layer, organizations may save time while removing the work through which human expertise was formed. The experienced professional remains valuable, but the path that produces the next experienced professional becomes harder to sustain.

The Geography Beneath the Cloud

Digital globalization encouraged the belief that geography was losing its importance. Software appeared weightless, information crossed borders instantly, and cloud services could be accessed from any connected location. AI seems to extend that movement by making advanced cognitive tools available through an ordinary screen.

Beneath that accessible surface lies an intensely physical system. AI depends on semiconductor fabrication, critical minerals, electricity, water, data centers, telecommunications networks, undersea cables, and sophisticated manufacturing equipment. Each component exists somewhere. Many are concentrated in a small number of companies, cities, and countries.

The more important AI becomes, the more strategically valuable those physical foundations become. Semiconductors are no longer treated as ordinary commercial products. Batteries, rare-earth processing, electrical grids, and data-center capacity have become matters of national planning. Governments support domestic industries, restrict the export of sensitive technologies, review foreign investments, and seek greater control over critical supply chains.

The cloud has geography, and that geography is becoming politically charged.

China demonstrates how far the map has moved from the early outsourcing narrative. It is no longer adequately described as a country receiving manufacturing work from developed economies. Chinese companies compete in electric vehicles, batteries, telecommunications, robotics, renewable energy, e-commerce, and artificial intelligence. The country has developed its own technological platforms and possesses enough economic power to shape international standards and supply chains.

Globalization has therefore become flatter and more fragmented at the same time. Individuals can communicate and collaborate across borders with unprecedented ease, but countries are constructing stronger boundaries around data, chips, energy, technology, and security. American and Chinese technology ecosystems increasingly reflect different corporate structures, regulatory priorities, and geopolitical interests.

The earlier hope that economic interdependence would reduce international conflict now appears incomplete. Interdependence can encourage cooperation, but it can also produce chokepoints. A country that controls a critical technology, material, shipping route, or manufacturing process possesses leverage over countries that depend on it. Sharing does not always dissolve power. Under certain conditions, it reorganizes power around the infrastructure that makes sharing possible.

Globalization was never a fully peaceful process. Its earlier forms included colonialism, exploitation, financial crises, labor displacement, and severe inequality. What has weakened is the post-Cold War confidence that greater trade and connectivity would draw countries toward a common economic and political order. AI is emerging in a world that is connected enough to depend on shared systems and divided enough to fear those dependencies.

Learning Faster Without Losing Judgment

For workers, the most common advice is that the contest will not be human versus AI, but human using AI versus human not using AI. This is a useful description of individual competition. A writer, analyst, engineer, or manager who understands how to work with AI may outperform a colleague who refuses to use it.

The formulation becomes less comforting when applied to an entire organization. One employee using AI may perform work previously distributed among several employees. If every worker becomes more productive, the organization may expand its output, reduce its workforce, or combine both responses. Universal AI literacy does not guarantee that every existing position will survive.

Even so, learning to use AI is becoming unavoidable for most knowledge workers. The required skill is broader than writing prompts. Prompting will become an ordinary form of computer literacy, comparable to using a search engine or spreadsheet. Durable value comes from knowing what to ask, recognizing whether the response is adequate, connecting it to a real organizational need, and correcting it when it fails.

Domain knowledge grows more valuable in this environment, not less. A cybersecurity analyst needs to understand threats well enough to challenge an AI-generated interpretation. A translator must recognize terminology, cultural meaning, and the consequences of an inaccurate rendering. A manager must know whether a proposed plan fits the people, constraints, and history of the organization. Without such knowledge, AI allows a person to produce plausible mistakes at greater speed.

My own workplace experience has revealed a smaller version of the uncertainty surrounding AI adoption. At one stage, employees were encouraged to use their available AI capacity extensively. Unused tokens appeared to represent missed opportunities for experimentation. Later, attention shifted toward controlling consumption and using resources more selectively. The organization moved from encouraging abundance to managing scarcity.

There is something comical about that reversal, but both phases reflect legitimate needs. Early experimentation helps people discover useful applications. Cost discipline becomes necessary when experimentation turns into routine operations. Trouble begins when token consumption is treated as a measure of innovation or token reduction is treated as proof of efficiency.

The more useful measure is the total cost of reaching a verified and usable result. A powerful model that completes a complex translation and layout task correctly in one attempt may cost less than a cheaper model that requires repeated correction. A costly model used to draft a routine message may provide no meaningful advantage. Employee time, review, delay, error, and reputational risk belong in the calculation alongside the price of the model.

Recent model development has made planning harder because capabilities and costs change faster than most corporate processes. An organization may spend months designing a workflow around one generation of AI, only to find that a newer model can perform the work through a different and shorter process. The arrival of systems such as ChatGPT-6 Astra reinforces the need for adaptable practices.

Adaptability should not become a demand to chase every product announcement. Some elements will change repeatedly: model selection, interfaces, token allocations, prompting methods, and agent architectures. Other foundations deserve greater stability: professional knowledge, verification standards, security, confidentiality, ethical responsibility, and understanding human needs.

AI literacy is non-negotiable, but AI dependence is not a strategy. People still need enough independent competence to identify an error, question an assumption, and continue working when a system is unavailable or inappropriate. The strongest combination is not a human surrendering thought to a machine. It is a human whose judgment gives direction to machine capability.

Organizations carry a related responsibility. They cannot purchase AI licenses, announce an upskilling program, and assume that adaptation will follow. Roles, training, incentives, quality controls, and career paths must be redesigned. If AI removes routine work, employers need new forms of apprenticeship that allow younger workers to learn through supervised evaluation, exception handling, and increasingly difficult decisions.

For the Philippines, the national challenge extends beyond training people to operate tools developed elsewhere. English proficiency and competitive salaries helped create the BPO industry, but those advantages will offer less protection when language itself is automated. Future strength will depend on specialized expertise, reliable infrastructure, secure data practices, global client relationships, locally developed intellectual property, and the ability to manage complete AI-enabled operations.

The Globalization of Intelligence

Should this new condition be called Globalization 3.5 or Globalization 4.0? If AI remained an improved form of software within the internet economy, 3.5 would be an appropriate description. Individuals would remain the central agents of globalization, supported by more capable digital tools.

The stronger term becomes justified when AI begins to perform tasks, coordinate processes, operate other software, and make limited decisions. The principal actor is no longer the unaided individual. It becomes a combined system of people, models, organizations, platforms, data, and computing infrastructure.

The World Economic Forum had already adopted the phrase Globalization 4.0 by 2019. Its definition addressed a broad transformation involving the Fourth Industrial Revolution, geopolitical change, inequality, environmental pressure, and the need for new institutions. Within Friedman’s progression, the term can be given a more focused meaning: the globalization of intelligence and agency.

Globalization 1.0 globalized the reach of countries. Globalization 2.0 globalized companies and production. Globalization 3.0 globalized the participation of individuals and the movement of knowledge work. Globalization 4.0 industrializes intelligence, making parts of it scalable, purchasable, and available on demand.

That development expands human capacity while placing pressure on human employment. It allows a Filipino professional to undertake work that once required a larger organization, but it may also allow a foreign company to obtain the same output without hiring that professional. The technology opens opportunity and removes opportunity through the same increase in productivity.

Ownership will determine how those effects are distributed. If a small number of companies control the models, chips, clouds, and platforms, global access to AI may coexist with concentrated economic power. The interface looks flat because millions of people can enter the same prompt box. The structure beneath it remains steep because only a limited number own the systems that respond.

Countries such as the Philippines must decide whether they will remain providers of affordable labor around technologies owned elsewhere or develop capabilities they can control. Moving upward requires more than placing AI beside existing processes. It requires building services, knowledge, institutions, and products whose value cannot be reduced to the number of inexpensive hours supplied.

Individuals face a related decision in their own work. Learning AI is necessary, but the purpose cannot be to imitate the machine or compete with it in speed. Human value will increasingly reside in framing the right problem, understanding consequences, managing relationships, evaluating uncertain evidence, accepting responsibility, and deciding what should be done.

The flat world once asked where work could be performed. The new flat world asks whether a human must perform it, who owns the intelligence that completes it, and how the people living with its consequences will share in its gains. Globalization 3.0 connected human intelligence across borders. Globalization 4.0 forces us to decide what kind of economy and society we will build when intelligence itself becomes global infrastructure.

Photo by Petr Macháček on Unsplash