How the AI arms race traps us all on an upgrade treadmill

From scams in Shenzhen to retraining courses in London, the accelerating pace of AI disruption is forcing everyone into a cycle of endless adaptation.

AI disruption is engineered as part of a "disruption-as-a-service" model, trapping societies in constant cycles of upskilling and insecurity. China offers a glimpse of a high-speed future, with open-sourced AI models and scams driving widespread anxiety and rapid adaptation. The burden of adaptation is shifting onto individuals, creating a universal treadmill where professional stability is in constant jeopardy.

The treadmill beneath our feet

The phrase AI arms race has become shorthand for geopolitical rivalry and corporate competition, yet it also describes a more intimate reality. Around the world, individuals find themselves pressed to learn faster, adapt harder, and spend more simply to remain relevant. Beijing community centres now run workshops for pensioners to spot deepfake calls, while charities in London organise scam-prevention courses for retirees. Meanwhile, professional translators in Shanghai or New York are racing to master neural models that threaten to automate their livelihoods.

This is not an accident. It is the hallmark of a new commercial logic: disruption as a service. Big Tech builds churn into the system, creating anxiety and obsolescence, then markets the very tools required to cope. The result is an upgrade treadmill, where the only option is relentless upskilling; with no guarantee that today’s investment will still matter tomorrow.

China’s glimpse of the high-speed future

China offers a vivid view of where this treadmill leads when technology, competition, and social anxiety converge. The stakes are not abstract. In February, Hong Kong’s branch of Arup, the British multinational, lost HK$200 million (US$25 million) when deepfake fraudsters impersonated senior executives in a video call. Meanwhile, viral scams have weaponised celebrity images such as diver Quan Hongchan, defrauding ordinary citizens out of their savings.

Such cases have heightened both public anxiety and regulatory urgency. But disruption is not limited to crime. It is baked into the workplace. Tencent’s decision to open-source two world-class AI platforms illustrates the paradox.

Hunyuan-MT, a translation tool that handles internet slang with uncanny precision, signals to translators that mastery of the tool is essential for survival. Hunyuan-Voyager, which can generate 3D worlds from a single image, poses existential questions for artists and game designers.

Democratising such models is a remarkable act of technological openness, but it also accelerates the treadmill. Professionals cannot afford to ignore them, nor can they assume mastery will secure long-term relevance.

A slower burn in the West

Western societies face similar tremors, though at a slower pace. Generative AI has sent enrolments surging on platforms such as Coursera, while organisations like Age UK now offer scam-prevention courses as a matter of public need.

Yet the treadmill is not merely consumer-driven. It is actively shaped by geopolitics. U.S. start-up Anthropic recently cut access for Chinese firms, a move with consequences far beyond corporate rivalry. Chinese companies that had integrated Anthropic’s models must now pivot to domestic or open-source alternatives. Intended as a strategic block, the decision has inadvertently fuelled the very arms race it sought to contain, forcing thousands of engineers and managers into abrupt technological recalibration.

The burden shifts to the individual

Perhaps the most striking feature of this treadmill is its individualisation of responsibility. A Pew Research Center survey found that more than half of U.S. workers worry about AI’s impact on their careers, with a third reporting they feel overwhelmed by the pace of change.

The commercial ecosystem compounds this anxiety. Meta has long faced criticism for enabling misinformation on its platforms. Yet it simultaneously offers AI literacy and misinformation detection tools — a neat illustration of the paradox. The harm is systemically created, while the cost of mitigation is borne by individuals.

For professionals, the new reality is relentless: scan Hugging Face for updates, test GitHub repos, enrol in the next course, or risk redundancy. What feels like self-improvement is often survival training against a system optimised to outpace us.

Rethinking the model

The treadmill metaphor is apt because it captures both the exertion and the futility. The more we run, the faster the machine seems to accelerate. Left unchecked, this dynamic risks entrenching not only digital exclusion but a deeper societal anxiety. The destabilisation of careers is no longer a marginal risk — it is becoming systemic.

The lesson is straightforward. Upskilling is necessary, but it cannot be the only answer. A shift in corporate ethos is overdue. Instead of disruption first, responsibility later, technology could be developed with security and human well-being at its foundation. This is not a utopian plea but a practical challenge to the current model. The arms race is not inevitable; it is a choice. And like any treadmill, we can decide to step off.

If the AI upgrade treadmill is built into the business model itself, how might Asian policymakers, investors, and technologists collaborate to design systems that prioritise stability as well as speed? The World Economic Forum has published research on the future of jobs and the need for reskilling in the age of AI.