Livelihoods based on knowledge work, also known as “white-collar jobs”, are competing with artificial intelligence. Addressing this growing socioeconomic tension should be a key priority at the 2026 APEC Finance Ministerial Meeting to be held in Hong Kong on Tuesday and Wednesday. The Silicon Valley narrative that AI will create unprecedented abundance should not blindly drive the pursuit of digital infrastructure at all costs, nor distract from the need for fiscal policies that protect citizens’ ability to earn a living.
Among APEC member economies, those with the highest measured knowledge-work intensity include Singapore (62 percent), the United States (52.5 percent), Australia (50.1 percent), Russia (47.8 percent) and Hong Kong (40.7 percent). Policymakers must prepare for an era of automated cognition, one where human intelligence is no longer essential for economic production. This is a new paradigm, and it is just getting started.
As of mid-2026, autonomous digital tools can effectively manipulate computer interfaces like keyboards, mice, and screens. They run on synthetic cognition that will soon outstrip even the most capable professionals. Most business leaders are only superficially aware of the situation. Enterprise consulting firms like Accenture and Boston Consulting Group are racing to enlighten them, even if this is increasingly seen as antisocial.
Consequently, much knowledge and clerical work that provides for modern livelihoods is fast becoming marginal or even unnecessary. Post-labor economists refer to this type of human labor as instrumental, only a means to an end. Customers and management value the final output rather than the human effort behind it.
Assuming senior management does not value human staff for their own sake, profit-driven companies will substitute them for technological automation when it crosses certain replacement thresholds. When the outputs are better, cheaper, faster and safer than what human employees can produce, they will switch. Unlike physical robotics that requires significant capital expenditure and factory retooling, autonomous software-using AI agents are cheap and frictionless. The only real barriers are the organizational redesign required to deploy them and the social acceptance of doing business with fewer people.
What looms is a massive upheaval in the relation between capital and labor. Based on the adoption curve for enterprise software, analysts predict widespread cognitive automation somewhere around 2030. Such timing is still plausible even if a financial bubble bursts in the meantime, which may even accelerate cheaper applications. We can expect that anything done on a computer now will be entirely or mostly automated. Far from being overhyped, the destruction of monetizable human knowledge work is likely underhyped.
Industry insiders have called for a pause. But AI development can’t stop and won’t stop. Private companies and governments are in a prisoner’s dilemma of acceleration, fearing that if one slows down, one’s competitors will not. Of the substitution thresholds, only safety remains a meaningful bottleneck. Cybersecurity fears may be the temporary savior of knowledge-based roles.
In 2027, companies will find AI solutions more accessible, more compelling, irresistible even. Simpler, more effective products will emerge. Experienced professionals will use them and multiply their productivity. But the total staff numbers will fall precipitously, and hiring will slow. Administrative support, customer service and legal review are highly exposed roles. Fresh graduates lacking domain experience are particularly vulnerable.
Despite the Hong Kong financial secretary’s stated priority for cultivating young financial talent, their digital, quantifiable work is among the ripest for automation. In the transactional world of finance, human labor is completely incidental to the desired output, a reality Hong Kong will need to adapt to.
Some livelihoods are more durable, rooted in essential demand for human presence: physical therapists, mental health counselors, elderly caregivers. Educators and doctors, even if more tech-enabled, may grow in number given their relationship with those they serve. Civil service jobs will inevitably become more automated, and so they should, to enhance quality and budget prudence. A prevailing counterargument is that AI will make everyone an entrepreneur. But, even with an army of AI agents, few human workers have the personal drive and the grit to succeed on their own in a competitive marketplace, especially one where everyone uses AI.
Policymakers must anticipate a near future with far less demand for discretionary knowledge work, and a citizenry with falling spending power. The APEC meeting presents Hong Kong with an opportunity to lead the needed rethink. Obviously, the appetite and preferences for solutions will vary across diverse political-economic systems. Foremost, it will be desirable to ensure productivity gains are not entirely captured as private economic rents, which aggravate inequality. New approaches to redistribution will be needed.
Disrupted livelihoods should be cushioned with mechanisms to expand household income beyond wages, which are likely to become compressed or lost. People will be more protected if a greater share of their income comes from asset ownership. Local wealth endowment funds could be created that compound and sustain better than universal basic income, especially against inflation and wealth consolidation. Companies could be encouraged through tax incentives to create shared equity trusts, which would be more socially useful than share buybacks.
Education and professional development should be reformed and expanded, with a focus on teaching effective AI usage to promote cognitive skilling and mitigate the cognitive regression that comes from overreliance on AI. Misuse of the technology is seductive and potentially catastrophic to human value-add.
Uncomfortable questions face APEC member economies. Should human livelihoods depend solely on maintaining a job that monetizes their labor? What mechanisms would allow all citizens of an artificially enabled state to share in the benefits? How can society be redesigned to facilitate a sense of purpose, fulfillment and productive contribution? Participating finance chiefs cannot assume traditional fiscal policies will suffice.
The author is the managing director of Global Institute For Tomorrow.
The views do not necessarily reflect those of China Daily.
