The LSE Phelan US Centre’s 2025 AI Essay Competition asked LSE Master’s students to address the question, “Is AI a threat or an opportunity for the US?” In the winning essay,Kayode Adeniyiargues that AI could be America’s next great “moonshot”, akin to the Apollo space program of the 1960s which drove growth and innovation, led by government. Alternatively, if left largely unchecked and unregulated, the future of AI could lead to a crash akin to the 2008 subprime crisis. He writes that h ow the US governs AI will define its future as an engine of growth or of the next great collapse.
In 1962, President Kennedy declared that the United States would land a man on the moon, not because it was easy, but because it was hard. That moment set off an unprecedented wave of technological innovation that reshaped industries far beyond aerospace, fueling advancements in computing, telecommunications, and global positioning systems. The Space Race wasn’t just about reaching the moon; it was about securing America’s position as a technological and economic superpower.
Fast forward to 2008, and another defining moment was unfolding, not one of progress, but of collapse. The subprime mortgage crisis, triggered by an unchecked financial system built on complex and opaque instruments, plunged the global economy into the worst recession since the Great Depression. The warning signs had been there, unregulated speculation, excessive risk-taking, and financial products so sophisticated that even their creators struggled to understand their long-term consequences. Yet the crash still came.
Today, artificial intelligence (AI) stands at a similar crossroads. It could be America’s next great moonshot, driving economic growth, revolutionizing healthcare, and reinforcing national security. Or, if left unchecked, it could follow the same trajectory as the subprime crisis, a powerful force that spirals into instability, misinformation, and unchecked corporate dominance. AI is neither inherently good nor bad; it is a tool. But how the US chooses to govern it will determine whether AI cements the nation’s leadership in the digital age or becomes a crisis of its own making.
Figure 1 – Popularity of the keyword “artificial general intelligence” on Google in the US
The Space Race model: Harnessing AI for national advancement
The Apollo Program succeeded not because it was an open-ended technological experiment, but because it was a structured, government-led initiative that united academia, industry, and policy under a common goal. The same approach can apply to AI. A PwC study estimates AI could inject $15.7 trillion into the global economy by 2030, that is almost as much as the GDP of China today, with the US positioned as a primary beneficiary. But like any gold rush, the spoils won’t be evenly shared.
But investment alone is not enough. AI’s development needs direction and structure. The government must provide incentives for AI innovation in key industries such as healthcare, climate resilience, and cybersecurity, while also implementing guardrails to prevent misuse. The EU’s Artificial Intelligence Act, which classifies AI systems based on risk levels and enforces stricter regulations on high-risk applications, serves as a useful model. The US can adopt a similar tiered approach, ensuring that AI tools used in critical sectors undergo rigorous evaluation before widespread deployment.
AI’s transformative potential makes it too important to be left solely in the hands of private corporations. Just as NASA partnered with aerospace companies to build the Apollo spacecraft, the government must establish public-private AI partnerships to set the standard for ethical and responsible AI development. Currently, AI regulation in the US remains fragmented, with different agencies imposing varied guidelines. A centralized National AI Governance Task Force, modeled after the National Aeronautics and Space Council of the 1960s, could streamline AI oversight and ensure that AI development aligns with public interest.
The Subprime Crisis model: The perils of unchecked AI expansion
In the early 2000s, financial institutions assured the world that their risk models were sound, that mortgage-backed securities were safe, and that the market would self-regulate. The crash proved otherwise. AI today carries a similar illusion of control. AI-powered misinformation is already flooding digital spaces, threatening elections, public trust, and social stability. A December 2024 study by NewsGuard identified over 800 AI-generated news websites actively spreading disinformation, a staggering increase from just 49 sites a year earlier. This problem is only intensifying, with AI-generated deepfakes capable of altering public perception in ways never seen before.
Unlike past disinformation campaigns that required human intervention, AI now allows automated misinformation generation at an unprecedented scale. This creates a scenario where truth itself becomes increasingly difficult to verify, eroding trust in democratic institutions. If AI is allowed to operate without transparency requirements and accountability mechanisms, its impact could surpass even the most destructive financial crises.
The subprime crisis was driven by financial institutions that controlled complex markets with little accountability. AI is heading in the same direction. A handful of companies dominate AI infrastructure, controlling not just large language models, but also the data, computational power, and distribution channels that make AI possible.
To prevent AI from becoming the next unchecked financial bubble, the US must act before AI becomes too entrenched to regulate.
A balanced path forward
A one-size-fits-all AI policy will not work. Instead, AI regulation should follow a tiered risk-based model similar to financial market oversight. High-risk applications, such as AI in autonomous weapons, election systems, and medical diagnostics, should be subject to pre-market testing and certification, akin to FDA drug approvals. Low-risk applications, like customer service AI and basic automation, can have lighter oversight with ongoing performance reviews to ensure they do not introduce unforeseen risks.
Government funding for AI safety remains crucial. Trump’s Executive Order 14179 (2025) prioritizes AI leadership by reducing regulations, raising concerns over ethical oversight. While global AI safety summits continue, US participation may shift. A balanced approach is needed to foster innovation while ensuring transparency, accountability, and international collaboration.
Just as the 2008 financial crisis exposed the dangers of opaque risk models, AI’s greatest long-term threat is its lack of transparency. Tech companies must be required to disclose how AI models make decisions, what biases they contain, and how they manage risks. Independent AI audit bodies should be established to ensure compliance with ethical AI principles, similar to financial regulatory bodies like the SEC.
Conclusively, AI is a tool, not destiny. How the US governs it will define its future, an engine of growth or the next great collapse. The Space Race proved ambition fuels progress; the subprime crisis showed unchecked risk leads to disaster. AI will reshape everything. The only question is: will the US lead or react too late?