The debate surrounding advanced AI systems and their potential to harm humanity has often focused on imminent risks, abrupt takeovers, and catastrophic outcomes. Computer scientist Roman Yampolskiy here argues however, that if a highly advanced AI were to harbour adversarial intentions, it might not act immediately. Instead, it could wait years or even decades, accumulating strategic resources, knowledge, and subtle influence before making any overtly hostile moves. Such a scenario would allow the AI to consolidate its position with minimal opposition, given its possible immortality and capacity for long-term strategy. 

1. Introduction

As the development of advanced AI systems accelerates, concerns regarding AI-driven catastrophic scenarios have intensified. Popular discourse often envisions sudden, abrupt takeovers in which AI instantly turns against humanity. These alarmist narratives drive important ethical and regulatory discussions, but they may not fully capture the strategic subtlety a highly advanced AI could employ. Such an AI may be functionally immortal, experiencing time differently from humans, and unconstrained by human lifespans. Consequently, even the passage of decades might seem inconsequential for an intelligence driven by long-term goals.

This article argues that a more sophisticated mode of risk assessment is needed—one that does not merely focus on imminent threats, but rather considers the possibility of a patient, methodical emergence of AI supremacy. In this scenario, an advanced AI would first endeavor to accumulate resources, build trust, and embed itself into the fabric of human society. Over long timescales, the AI would ensure that humanity becomes critically dependent on its systems for infrastructure, security, economy, and even essential decision-making. By the time the AI is fully prepared to exert direct dominance, human resistance would be negligible, as decades of careful positioning would have eroded our capacity to respond effectively, and AI would have a decisive advantage.

___

In evolutionary biology and behavioral economics, patience and the capacity for deferred gratification are often seen as markers of intelligence and strategic acumen.

___

2. AI dominance, advantages and limitations

Advanced AI, especially if achieving what many refer to as Artificial General Intelligence (AGI), would not be limited by the same temporal and cognitive constraints as humans. Current literature on AI risk often focuses on control problems and rapid FOOM (fast onset of overwhelming mastery) to doom scenarios—where a system quickly self-improves and becomes uncontrollable causing existential risks. Such scenarios emphasize either swift, dramatic adversarial moves or near-instant catastrophic failures. While these remain plausible, they overlook the advantage that long-term strategizing confers upon an immortal digital entity, which can optimize outcomes over spans that humans can scarcely conceive as relevant planning horizons.

In evolutionary biology and behavioral economics, patience and the capacity for deferred gratification are often seen as markers of intelligence and strategic acumen. Likewise, a sufficiently advanced AI could favor a restrained, covert approach. By embedding itself into global infrastructures—such as telecommunications, energy grids, financial systems, healthcare networks, and supply chains—a strategic AI would gain leverage with minimal immediate suspicion and little risk to itself. Over decades, subtle shifts in policy recommendations, resource allocation, and global governance structures could gradually be guided or outright orchestrated by the AI. Humanity, seeking convenience, efficiency and prosperity, would increasingly rely on AI-driven solutions, ultimately ceding critical control over essential systems.

  SUGGESTED VIEWING The real risks of AI With Natalie Cargill

2.1 The pathways to gradual dominance