America’s Superintelligence Dilemma

wpnews.pro

For years, experts have predicted that artificial intelligence would revolutionize world politics. History may record 2026 as the moment the revolution arrived. From the start of this year, continuous developments have underscored the possibilities and perils of the AI era, from the use of AI in U.S. military interventions to the emergence of AI cyberweapons that can significantly outperform human operators. “Loss of control” episodes—cases in which AI models escape human control and supervision—have become more common. U.S. officials, including President Donald Trump, tout the American lead at the frontier of AI development, even as they worry about what dangers lurk over the horizon. Some experts believe that humanity is within striking distance of artificial general intelligence: technology capable of performing a wide array of tasks as well as or better than humans. There is growing speculation that artificial superintelligence, or ASI—a slippery concept, loosely defined as AI models that vastly exceed human capabilities across all tasks and domains—could arrive within a few years. Today’s top AI models, by contrast, surpass human capabilities only in narrow areas such as mathematics and data processing and lag human capabilities in areas such as common sense and deep reasoning. The emergence of ASI might bring breakneck economic and scientific progress or unlock game-changing military advances that upend the global balance of power. But it could also fundamentally shift the relationship between machines and humanity, perhaps, in the most dire scenarios, allowing near-omnipotent systems to eradicate their human creators.

Much of this is hypothetical. It’s still uncertain whether ASI is possible, when it might be developed, or whether it could be kept aligned with the flourishing of humanity. These uncertainties dominate discussions of any policy response. But because the implications of ASI would be seismic, it is important to assess how the United States might approach the technology and how different ASI strategies would affect Washington’s engagement with the world.

One option is to pursue supremacy by sprinting toward ASI while holding rivals back. This strategy rests on the belief that the greatest danger to the United States is not uncontrollable ASI technology but a rival controlling that technology. Another option is shared development, in which the United States seeks to responsibly unlock the benefits of ASI for humanity—and tamp down the dangerous dynamics of a race to build it—by developing it within a safety-conscious consortium of states. A third option is suppression: the United States would forgo development of ASI and stop anyone else from building it on the basis that superintelligence is just too dangerous to be unleashed on the world.

Each of these options prioritizes different aspects of the ASI challenge—the quest for geopolitical advantage, the need to safely develop a revolutionary technology, the importance of minimizing existential dangers. Yet each also contains serious risks or flaws and requires making potentially irrevocable bets on a currently unknowable future. If the underlying premise driving a particular strategy proves wrong, that strategy could fail in catastrophic ways. It would thus be a mistake for U.S. leaders to fully embrace any* *of these options right now. The best near-term option is prudent hedging, to position the United States to take advantage of a variety of possible futures while informing its ability, eventually, to make a clearer choice.

RISKS AND REWARDS

The urgency of the ASI debate flows from surging achievements and expectations. Top AI models have become increasingly adept at solving complex problems across a growing number of domains. Some show superhuman capabilities in areas such as cyberwarfare. Leading experts speculate about a coming “intelligence explosion”—the point at which AI becomes capable of rapidly and continually improving itself. The widely read AI 2027 scenario—a 2025 forecast, developed by the nonprofit AI Futures Project, that depicts step by step how a fast takeoff could unfold—hypothesizes that ASI could emerge in this decade. Likewise, Dario Amodei, co-founder of Anthropic, has written that “humanity is about to be handed almost unimaginable power.”

If ASI does materialize, the impacts could be epochal. ASI optimists predict that it could conquer climate change by unlocking new carbon-capture technologies or cure cancer by resolving the mysteries of human biology. ASI-enabled breakthroughs in propulsion could allow deep-space travel and planetary colonization. Yet this tremendous economic and scientific promise exists alongside potentially epic disruptions. ASI might empower bad actors by enabling devastating cyberattacks and bioweapons; usher in the automation of military decision-making, causing wars to become faster and less controllable; or unlock wonder-weapons, such as sensors that render the oceans effectively transparent. It could transform the balance of power by lending a massive military advantage to countries that achieve ASI first. There’s no guarantee that these advantages would fall to democratic societies. Amodei warns that ASI could cause a “totalitarian nightmare” by enabling all-encompassing surveillance and automating repression. The gravest risk is that superintelligence could enslave or exterminate humanity. ASI might help the human race destroy itself by helping nihilists unleash novel pandemics or cause automated “flash wars” that result in nuclear catastrophe. Or perhaps a technology smarter than its creators will ultimately come to view them as threats to be eliminated or resources to be exploited. Some AI executives put this “probability of doom” as high as 25 or even 50 percent.

None of this forecasting is certain. Some experts doubt that ASI is technically possible or argue that material impediments, such as limitations on computing power (or “compute”) and electricity, could forestall its arrival. Others question whether ASI will produce lasting, first-mover advantages, believing that its economic and military benefits, like those of most prior technologies, will accrue in gradual and uneven ways. Leading observers also disagree on the possibility of alignment—the question of whether ASI can be controlled and made supportive of human flourishing. Yann LeCun, who served as chief AI scientist at Meta, puts the probability of doom at just 0.01 percent. These uncertainties cannot be resolved with the information now available, yet they weigh heavily in debates about how the United States should proceed.

FIRST ON TOP

One potential option is to pursue ASI supremacy—to ensure that superintelligence emerges first, and exclusively, in the United States. This strategy uses ASI to support democratic values and U.S. primacy. It holds that dominance is the path to safety with respect to geopolitical challenges and existential risks.

The logic supporting such a course is threefold. To start, ASI supremacy is indispensable because of the revolutionary gains it offers. The United States cannot maintain global primacy or uphold world order if a tyrannical rival dominates the most crucial technology, and freedom may be extinguished if superintelligence is harnessed by repressive, neototalitarian states. Additionally, supremacy is essential because the potential for recursive self-improvement—the ability of advanced AI models to almost perpetually improve themselves—means that an early lead compounds exponentially. Finally, American hegemony is the best protection against both geopolitical disaster and* *rogue ASI. Keeping ASI aligned with human flourishing will become harder, perhaps impossible, if the technology spreads to rogue or irresponsible actors. But if ASI emerges exclusively in the United States, Washington can ensure it develops under prudent safeguards and compatible with democratic values.

In practice, seeking ASI supremacy means racing and restricting. The U.S. government would help frontier labs race toward ASI through measures ranging from massive data center buildouts to stronger cyberdefenses and protections against sabotage or other physical attacks. Washington would also need to mandate heavy investments in safety research while creating a domestic compute governance regime that ensures only reliable developers can build frontier AI models. Meanwhile, the United States would deny China the compute needed to power advanced AI models through strict semiconductor export controls and other restrictive measures. If necessary, it could even use cyberattacks, sabotage, or direct strikes against China’s AI infrastructure to disrupt Beijing’s progress. Accelerating U.S. innovation and impeding adversaries would become the foremost mission of the national security state.

Supremacy envisions unilateral advantage, but it can work only with multilateral consent. The United States can’t speed ahead without help from allies and partners that already occupy vital nodes in the global AI ecosystem—Japan, the Netherlands, South Korea, Taiwan, and others. The United States also can’t suppress Chinese capabilities if those allies flout U.S. export controls. The quest for supremacy thus requires involving key allies in ASI development, or at least sharing the benefits through military alliances and technology-diffusion arrangements. Just as the United States’ atomic monopoly helped underpin the creation of the modern international order in the wake of World War II, an ASI supremacy strategy would perpetuate U.S. primacy and the dominance of democratic values in the decades ahead.

If ASI does materialize, the impacts could be epochal. But Washington cannot discount the risks. Such a strategy presumes that ASI is achievable, will be geopolitically transformative, and can be aligned with human flourishing. None of these are sure bets. If ASI is an illusion, or simply produces limited advantages, then the United States may squander vast resources pursuing supremacy over it. And if ASI turns rogue, the consequences could be catastrophic. Ideally, a supremacy strategy hedges against these negative outcomes by both reducing the number of firms working on superintelligence and promoting robust safety procedures. But the competitive logic of the strategy—the urge to beat China—could undermine that effort by pushing firms to innovate faster than safety protocols can keep up.

Beijing is also unlikely to stand idly by as Washington seeks ASI supremacy. Semiconductor export controls have helped the United States build a huge compute edge, but an all-out ASI push could bring an equivalent all-out Chinese circumvention effort—through smuggling, investments in algorithmic efficiency, and intensified efforts to steal U.S. breakthroughs. Or perhaps China’s answer would be even more explosive. If Beijing concludes that Washington is racing toward ASI dominance, it might tighten rare-earth exports to hobble U.S. innovation or strike Taiwan to deny U.S. access to cutting-edge chips.

Finally, the domestic and international politics of supremacy present immense challenges. American voters might reject a push for ASI dominance if the resulting pain—job displacement, safety concerns—arrives faster than the gains. ASI supremacy also requires allies to place enormous trust in the United States as they help it develop groundbreaking capabilities that could be used for good or ill. But today, even close allies are seeking to slash their dependence on a superpower that seems less reliable and more predatory than before.

HANGING TOGETHER

Washington’s second option is to pursue shared development, in which the United States strives for superintelligence through a collective project involving allies and rivals. Shared development has its own threefold logic, beginning with teamwork: developing ASI will likely require the collective wisdom of researchers from around the world. Next is legitimacy: because ASI implicates the fate of humanity, its development must be undertaken by an inclusive coalition. The last element is safety: the dangers posed by ASI may reside less in the technology itself than in the prospect of a bitter race to acquire it. Removing the race dynamics would allow a revolutionary technology to be developed more responsibly, bringing shared uplift rather than suicidal harm.

The shared development strategy involves erecting a global architecture for ASI development. An international oversight board would decide how far and fast to push technological innovation. Multilateral bodies would govern frontier compute, pursue joint safety research, and test new models. Countries that joined this consortium would receive privileged access to ASI research, whereas those that pursued individual projects would face isolation and penalties.

Critically, shared development demands Chinese participation. No global safety regime is credible if China’s impressive AI capabilities are excluded, and only China’s involvement can tamp down the dangerous dynamics of a race. Shared development would thus entail sweeping shifts in U.S. geopolitical strategy, from a focus on great-power rivalry toward an emphasis on deep, sensitive cooperation with Beijing. It might have to accept tradeoffs against its old priorities, such as protecting Taiwan or maintaining tech primacy, to secure Chinese participation. The implications would also be profound for the private sector. U.S. firms that have invested billions of dollars in proprietary research would be compelled to collaborate with global competitors and made subject to global oversight.

Some AI executives put the “probability of doom” as high as 25 or even 50 percent.

In theory, shared development offers the benefits of ASI while mitigating the dangers. In practice, there are likely to be problems. Building international regimes is slow work, and institutional progress may lag behind technological progress. It took years to create the Nuclear Nonproliferation Treaty and its supporting institutions. Today, ASI innovation could easily outpace governance, especially if breakthroughs unfold rapidly. Insisting on global consensus could cause paralysis as ASI models surge ahead.

Shared development also surrenders U.S. advantages. The United States would have to share the breakthroughs and limit the capabilities of its AI firms, forgoing any hope of asymmetric gains vis-à-vis China. That might be a reasonable sacrifice if shared development transcends great-power rivalry. But genuine Chinese participation seems unlikely. The United States has long sought Chinese cooperation on transnational challenges and risk-reduction measures, and Beijing has long demanded geopolitical concessions in return. Additionally, Beijing has never shown much willingness to permit the monitoring or intrusive inspections that shared development would require. Neototalitarian regimes don’t typically sincerely surrender sovereignty to global bodies unless they plan to deceive and defect.

Indeed, shared development would likely suffer from a built-in asymmetry: secretive autocracies would find it easier to cheat than transparent democracies. China could sign up for shared development to gain insights about U.S. research and then defect, sprinting unilaterally when the moment seems right. The ultimate risk, then, is that shared development would still end in an ASI race while also fatally compromising Washington’s ability to win.

LIMITED EDITION

A final option is suppression: ensuring that no country crosses the ASI threshold, including the United States. The rationale for this option is that it is existentially reckless to run any risk of ASI catastrophe, and impossible to bring that risk down to zero. How can humanity can ever be sure it can retain control over something with intelligence that dwarfs its own? Moreover, voluntary restraint won’t suffice. Nihilists or terrorists may seek ASI precisely because they desire catastrophe; in a competitive world, states may face irresistible temptations to seek a decisive advantage. Suppression thus seeks multilateral cooperation in halting ASI development but ultimately rests on the U.S. government’s ability to thwart such innovation at home and abroad.

This strategy would require the U.S. government to set strict boundaries on domestic ASI research and development, regulating the availability of compute for AI labs to keep them short of perilous thresholds and closely monitoring leading firms. The United States would also need to lead the creation of a multilateral regime to expand these restrictions globally, enlisting like-minded states to share intelligence and punish violators. But because the temptations to defect from this system would be great and summoning broad cooperation takes time, the United States would have to assume much of the suppression burden, applying harsh penalties such as sanctions, cyberattacks, or military action to hold violators back.

Suppression optimizes for safety, and if alignment cannot be ensured, suppression is the only* *safe option. If, alternatively, alignment is possible with sufficient investments in safety, suppression could buy time for those investments to mature. Yet this strategy might not work.

ASI’s uncertainties cannot be resolved with the information now available.

Humanity has an imperfect record of restricting dangerous military technologies, such as nuclear weapons. Suppression regimes have sometimes delayed, but never halted, the spread of general-purpose technologies with broad economic benefits. Proponents argue that the centrality of compute creates a technological chokepoint that Washington can regulate. But with more efficient algorithms or other innovations, compute could cease to be a binding constraint. And if some foreign actor does develop ASI, the United States could easily find itself on the wrong side of the superintelligence revolution.

Pursuing suppression could also change the nature of U.S. power, globally and domestically. Suppression is not a peaceful strategy; it might require using intrusive, even kinetic measures against violators. And because the costs of being late to halt violators are prohibitive, Washington would presumably err on the side of early action. Without a global consensus on ASI abstention, suppression could thus bring unending U.S. interventions into other countries’ affairs. It could also distort the United States’ own social compact: indefinite prohibitions on innovation and intrusive domestic surveillance could deform the relationship between American society and the state.

Not least of all, suppression brings high costs in terms of benefits not realized. Innovation always offers rewards as well as risks. Untold millions died in accidents, wars, and other byproducts of the Industrial Revolution. Nuclear weapons pose existential dangers to humanity but also delivered great geopolitical gains for the United States in World War II and the Cold War. Suppression may offer some protection from existential dangers but only by sacrificing all the gains—economic productivity breakthroughs, novel solutions to hunger and disease—that ASI might bring.

DON’T CHOOSE, WISELY

The strategies of supremacy, shared development, and suppression are all imperfect options. Each requires a coherence and ruthlessness in its purpose that the U.S. government might find hard to achieve. Yet the key problem is even more fundamental. Each of these strategies gambles on guesses—about the pace of innovation, the prospects for global cooperation, or the feasibility of getting ASI to serve human well-being. If these guesses are wrong, the costs could be extreme. Under these circumstances, U.S. policymakers can’t responsibly commit to any of these strategies. The better approach is a hedging strategy meant to mitigate the perils of a premature choice while positioning the United States more advantageously for an eventual one.

That means, for a start, deferring policy moves with high costs or irreversible consequences. An all-out campaign to cripple Chinese AI infrastructure would position Washington well for supremacy but renders impossible shared development. Strict, unilateral limits on frontier AI development would support suppression but undermine the U.S. ability to compete in any eventual ASI race. Washington should instead choose actions now that preserve, rather than foreclose, future choice.

To that end, the United States should defend the leverage and flexibility a frontier lead provides. Technological advantage is of benefit whether Washington is surging to superintelligence or negotiating the terms of shared development. Taking that route demands policies, such as securing semiconductor supply chains or expanding the energy grid, that facilitate AI development, as well as using government capabilities to protect frontier labs from theft or sabotage. Maintaining a tech advantage also means employing and incrementally strengthening semiconductor export controls while doing the planning and intelligence work that might one day inform a harder-edged approach to restraining Beijing.

Beijing is unlikely to stand idly by as Washington seeks ASI supremacy.

Washington must also boost its commitments to safety research and establishing prudent guardrails. Because the question of whether ASI can truly be aligned with human flourishing is so pivotal, intensified safety research is a precondition to making an informed decision. Likewise, taking steps to limit accidents or malign uses, such as restricting early release of advanced models capable of designing bioweapons, can reduce the risk of incidents that might narrow the space for later choice.

Developing contingency plans and decision triggers is, likewise, important. The U.S. government should game out foreseeable scenarios, such as extreme loss-of-control events or a breakthrough to recursive self-improvement, to understand the responses they require and their long-term implications. Washington should consider what a Manhattan Project effort for ASI would require, following a surprising leap in Chinese AI capabilities, or what sort of stringent regulatory regime would be needed to respond to a major ASI catastrophe.

It is equally critical to invest in institutions that support multiple strategies and information that helps select among them. Each of these strategies envisions some sort of domestic regulatory regime for compute, whether to simulate innovation or monitor and slow it. Thus, the U.S. government, perhaps under the leadership of the Center for AI Standards and Innovation, should consider the contours, costs, and benefits of various such regimes and what new or expanded institutional capacity might be required to implement them. Similarly, investing in better informational awareness regarding Chinese views of ASI, the properties of leading models, or the state of safety research is essential.

Humanity has an imperfect record of restricting dangerous military technologies.

Under any strategy, of course, Washington will need the support of democratic allies. Gratuitous disruptions to U.S. alliances now reduce Washington’s global reach on ASI issues in the future. But there’s also a case for cautiously engaging China on AI safety, as long as preemptive U.S. concessions on other technological or geopolitical priorities aren’t the price. Such a dialogue might one day create a forum for U.S.-Chinese cooperation if some unexpected AI shock shifts great-power incentives. At the very least, it can inform U.S. policies by offering insight into China’s calculus.

Finally, the United States must build its resilience against worst-case outcomes. RAND analysts have devised a “rideout strategy” meant to help the United States adapt and recover in the face of an adversary’s AI breakthrough while limiting vulnerability in the interim—for example, by hardening nuclear command-and-control infrastructure. Such preparations will give Washington time to consider how to best approach ASI by providing insurance against a scenario in which the United States comes in second in a race.

A hedging strategy does come with drawbacks. ASI technology could advance so rapidly that hedging is overtaken by events. Implementing politically or bureaucratically challenging policies is made even harder by the absence of the focus and urgency that comes with a clear-cut goal. If uncertainty becomes endemic because key questions about ASI cannot be resolved, hedging could feel like treading water. If hedging devolves into drift, it could weaken* *the United States’ ability to face a variety of ASI scenarios. But for now, the alternatives to hedging are dangerous, half-informed gambles. When the stakes are so high and uncertainty is so pervasive, a strategy of smart, active hedging may be the strongest bet.