Geopolitical Feasibility of the AI Futures Project's Plan A for Delaying Superintelligence

The AI Futures Project recently published AI 2040: Plan A on lessw-blog, outlining a normative policy scenario to delay artificial superintelligence development from a projected 2030 to 2040. PSEEDR analyzes the geopolitical feasibility of this proposed US-China regulatory alignment, contrasting the framework's coordinated intervention with the current reality of escalating semiconductor export controls and tech-nationalism.

The AI Futures Project recently published "AI 2040: Plan A" on lessw-blog, outlining a normative policy scenario to delay artificial superintelligence (ASI) development from a projected 2030 to 2040. PSEEDR analyzes the geopolitical feasibility of this proposed US-China regulatory alignment, contrasting the framework's coordinated intervention with the current reality of escalating semiconductor export controls and tech-nationalism.

The Architecture of Plan A

The AI Futures Project's latest scenario represents a significant methodological shift in AI safety literature. Rather than forecasting a probabilistic future, the authors present a normative roadmap-a detailed, step-by-step policy plan illustrating what should happen to safely manage the transition to artificial superintelligence. The core premise of the source material asserts that without decisive intervention, ASI is on track for realization by 2030. "Plan A" posits that coordinated action between the United States and Chinese governments could successfully delay this threshold to 2040, providing a crucial decade for safety research and alignment frameworks to mature.

By hosting the complete scenario at a dedicated domain, the project builds upon its previous "AI 2027" methodology, emphasizing comprehensive planning over high-level abstractions of existential doom or utopian acceleration. The authors explicitly state that this is a recommendation of what should happen, acknowledging the immense difficulty of execution while maintaining that the path is plausible enough to warrant serious policy consideration.

The Geopolitical Feasibility of US-China Coordination

The linchpin of "Plan A" is bilateral cooperation between the world's two leading AI superpowers. However, PSEEDR's analysis indicates a stark contrast between this idealistic regulatory alignment and the current geopolitical reality. The prevailing paradigm is defined by intense tech-nationalism and aggressive decoupling strategies. The United States has implemented sweeping semiconductor export controls, utilizing the Entity List and the CHIPS Act to restrict China's access to advanced GPUs and the extreme ultraviolet (EUV) lithography equipment required to fabricate them. In response, China has accelerated its domestic silicon initiatives and restricted exports of critical minerals essential for semiconductor manufacturing.

Achieving the coordination envisioned in "Plan A" would require a monumental diplomatic pivot, akin to the Cold War-era nuclear non-proliferation treaties, but applied to a dual-use technology that is deeply integrated into both economic competitiveness and national security. The feasibility of such an alignment hinges on both nations mutually recognizing ASI as a shared existential threat that supersedes zero-sum economic and military advantages. Currently, the incentive structures in both Washington and Beijing heavily favor rapid capability scaling over unilateral or bilateral deceleration. The assumption that both states would willingly throttle their most strategic technological sector requires a fundamental realignment of global power dynamics, making the political capital required for "Plan A" exceptionally difficult to mobilize.

Implications for the AI Safety Discourse

Despite the geopolitical friction, the publication of "Plan A" carries significant implications for the broader AI safety and governance ecosystem. Historically, the AI safety discourse has been dominated by abstract philosophical arguments regarding existential risk, orthogonality, and instrumental convergence. While theoretically valuable, these abstractions often fail to translate into actionable policy. "Plan A" forces the conversation into the realm of concrete, timeline-based policy planning.

By establishing a specific target of delaying ASI to 2040 and outlining the state-level interventions required to achieve it, the AI Futures Project provides a tangible framework for policymakers, researchers, and regulatory bodies to critique and iterate upon. This shift is critical. It moves the Overton window from debating whether ASI is a risk to debating the specific mechanisms of state intervention required to manage artificial general intelligence timelines. It also highlights the reality that private corporate governance is insufficient; managing a technology of this magnitude requires decisive state capacity and international diplomacy. For frontier AI labs, this signals a potential future where state intervention overrides internal corporate safety boards and self-imposed scaling caps.

Limitations and Open Questions

While "Plan A" offers a structured approach to ASI delay, several critical limitations and open questions remain unaddressed in the foundational brief. First, the scenario lacks granular detail on the specific regulatory mechanisms and enforcement tools required to ensure US-China bilateral compliance. Verification in software and algorithmic development is notoriously difficult compared to counting nuclear warheads or monitoring uranium enrichment. A compute-cap treaty, for example, would require unprecedented inspection access to sovereign data centers, a concession neither nation is currently prepared to make.

Second, the proposal requires a rigorous technical definition and universally accepted metrics to identify the threshold of superintelligence. Without objective benchmarks-such as specific loss thresholds, autonomous replication capabilities, or cross-domain reasoning metrics-enforcing a delay becomes a moving target subject to political interpretation. Finally, the framework must address the high probability of defection. Even if the US and China successfully coordinate, the proliferation of open-weight models and the decreasing cost of compute create vectors for non-state actors, private entities, or third-party nations to bypass bilateral restrictions. Tracking decentralized compute clusters poses a fundamentally different challenge than monitoring centralized hyperscalers.

Synthesis

The AI Futures Project's "Plan A" serves as a necessary forcing function for AI governance, demanding that the industry move beyond passive forecasting and embrace active, normative policy design. While the geopolitical hurdles to US-China coordination are immense, characterized by entrenched tech-nationalism and hardware embargoes, the exercise of mapping a managed timeline to 2040 provides a vital blueprint for international diplomacy. Ultimately, the success of any such framework will depend not just on bilateral agreements, but on the development of robust, verifiable enforcement mechanisms capable of surviving the intense competitive pressures of the global AI arms race.

Key Takeaways

• The AI Futures Project proposes a normative policy plan to delay artificial superintelligence to 2040 via US-China coordination.
• Current geopolitical realities, including semiconductor export controls and tech-nationalism, present significant friction to the proposed bilateral alignment.
• The framework successfully shifts AI safety discourse from abstract existential risk to concrete, timeline-based policy planning.
• Significant open questions remain regarding enforcement mechanisms, technical definitions of ASI, and the risk of defection by non-state actors.