AI’s dual nature: Transformative force or irreversible threat?

The study warns that this proliferation creates a governance dilemma: while treaties and international agreements once helped regulate nuclear threats, AI evolves at “software speed,” rendering such mechanisms inadequate. Safety and accountability must be engineered into AI systems through built-in evaluation, provenance tracking, compute monitoring, and third-party audits. The governance framework must evolve from bureaucratic oversight to agile, interoperable mechanisms that match the speed of digital innovation.

Artificial intelligence stands at a crossroads between historic continuity and existential rupture, argues a new study by Masoud Makrehchi of Ontario Tech University. The research presents a multidimensional framework for understanding AI’s evolution and its governance needs, framing it as both a transformative economic force and a potential civilization-scale risk.

Published as “Three Lenses on the AI Revolution: Risk, Transformation, Continuity” on arXiv in October 2025, the study dissects AI’s global trajectory through three analytical perspectives, risk, transformation, and continuity, asserting that artificial intelligence is simultaneously evolutionary and revolutionary. It predicts that AI will drive productivity and reshape labor, while also carrying irreversible consequences that demand urgent governance and ethical foresight.

The risk lens: AI’s nuclear paradox

The first of Makrehchi’s lenses situates AI alongside nuclear technology, not in physics but in societal consequence. The study argues that both domains carry tail risks, rare but catastrophic outcomes that could be irreversible. While nuclear technologies are bound by material scarcity and complex infrastructure, AI’s diffusion is unconstrained, its barriers to entry drastically lower. The same models that accelerate drug discovery or improve logistics can also be weaponized for cyberwarfare, automated propaganda, or biosecurity breaches.

The study warns that this proliferation creates a governance dilemma: while treaties and international agreements once helped regulate nuclear threats, AI evolves at “software speed,” rendering such mechanisms inadequate. Safety and accountability must be engineered into AI systems through built-in evaluation, provenance tracking, compute monitoring, and third-party audits. The governance framework must evolve from bureaucratic oversight to agile, interoperable mechanisms that match the speed of digital innovation.

Another key concern is opacity. Advanced AI models, particularly large-scale generative systems, operate as black boxes, producing outcomes that even their developers struggle to explain. This unpredictability introduces unknown risks in causal reasoning, bias propagation, and decision-making autonomy. The research emphasizes the need for red-teaming, continuous testing, and rigorous incident reporting as baseline governance strategies. Without them, societies risk crossing one-way thresholds, irreversible decisions delegated to unaligned or unaccountable systems.

The paper argues that AI’s risks are not evenly distributed but globalized. A single design flaw or malicious model release can spread worldwide in hours, outpacing the ability of regulators to respond. Such acceleration creates what Makrehchi calls “singularity-class tail risks,” scenarios where small-scale technical failures could escalate into systemic crises. This nuclear-like dimension underscores AI’s dual-use nature—technologies of immense promise paired with threats of equal magnitude.

The transformation lens: Industrial lessons for a cognitive economy

While AI’s risks are existential, its transformative power is undeniable. Through its second lens, the study equates AI to the Industrial Revolution, describing it as a general-purpose technology that automates cognition rather than muscle. Just as mechanization and electrification redefined labor in the 18th and 19th centuries, AI extends automation to reasoning, analysis, and decision-making—tasks once thought uniquely human.

Makrehchi observes that AI’s economic transformation is skill-biased rather than job-destructive. Instead of eliminating work wholesale, AI redistributes it, automating tasks within roles and elevating demand for human judgment, creativity, and cross-domain integration. The emerging economy prizes abilities that machines cannot replicate: ethical reasoning, contextual understanding, and trust-based interaction.

The research draws several analogies with industrialization. First, automation and scaling pressures are reshaping modern organizations. Just as factories reorganized production around machinery, companies today are restructuring workflows to be “AI-first.” Second, AI exerts deflationary pressure on cognitive services. The marginal cost of translation, summarization, and drafting now approaches zero, mirroring the price collapse in mass manufacturing. Third, standardization of quality, once achieved through interchangeable parts, is now enforced through model benchmarks, automated evaluation suites, and digital guardrails.

The study also identifies a “barbell effect” in knowledge work. Routine analysis and mid-tier drafting are rapidly commoditized, while bespoke, high-trust services, such as legal strategy or brand-specific creative direction, command premium value. This bifurcation mirrors the industrial divide between mass-produced goods and artisanal craftsmanship.

However, transformation brings turbulence. The transition demands rapid reskilling and institutional adaptation to prevent displacement and inequality. Policymakers, the study argues, must embed competition policies at platform layers, invest in AI literacy, and treat computational infrastructure as public capital. Sustainability must also be prioritized: training large AI models consumes immense energy and water, potentially locking in environmental costs comparable to industrial pollution.

AI’s governance challenges thus echo those of early industrial societies, balancing innovation with social stability, concentration of power with equitable access, and economic expansion with ecological responsibility. The author concludes that the outcomes of this revolution will hinge not on algorithms but on the institutions that manage them.

The continuity lens: The fourth computing revolution

The third interpretive frame places AI within a half-century arc of technological evolution, from personal computing to the internet, mobile communication, and now intelligent automation. Each of these revolutions, Makrehchi argues, expanded both the scope of automation and the circle of access, pushing technology closer to the individual user.

In this view, AI represents continuity rather than rupture. It extends the democratization of digital power while reinforcing structural patterns familiar from earlier eras: concentration of production, rapid adoption, and falling marginal costs. The study traces this lineage across four waves. Personal computers democratized data processing; the internet democratized information sharing; mobile networks democratized connectivity; and AI now democratizes cognition itself.

This sequence reveals recurring social and economic dynamics. Each revolution produced power-law markets, with a few dominant firms and a long tail of smaller players. Each saw open-source movements act as counterweights to commercial monopolies. Each also deepened the trade-off between privacy and convenience - a pattern now intensified as AI systems learn from personal data to anticipate user intent.

AI’s defining distinction is its speed. Whereas personal computing took decades to reach global penetration, generative AI applications achieved mass adoption within months. This acceleration amplifies both benefits and risks: democratization of knowledge coexists with concentration of control. The study cautions that without deliberate design, personalization can become an attack surface, enabling manipulation, surveillance, and algorithmic bias at unprecedented scale.

The paper notes that as machine-generated content proliferates, human value shifts from production to verification. Writers become system designers, readers become auditors, and originality becomes the last refuge of human distinctiveness. In this “verification-first” information economy, provenance, trust, and accountability are the new currencies of credibility.