A new economic model from the University of Chicago shows that the race to artificial general intelligence (AGI) isn't just a technical problem; it's a market problem. When companies compete to build AGI first, they face a fundamental trade-off: invest resources in speed to reach the finish line, or invest in safety to reduce catastrophic risk. The research reveals that market structure itself can push firms toward riskier choices, even when those choices harm everyone.

How Does Competition Change AI Safety Incentives?

Researchers at the Becker Friedman Institute studied a model where firms must allocate scarce resources between two competing goals. Speed increases a company's chance of reaching AGI first, but it leaves fewer resources for safety measures. Safety reduces the risk of catastrophic outcomes, but it slows development. The key finding: fragmentation of the AI industry increases both total speed and conditional doom risk by shifting resources away from safety and toward speed.

The model identifies a critical market size threshold. Below a certain threshold, firms have positive expected payoff from achieving AGI, so they race. Above that threshold, firms race even though achieving AGI has negative expected value for society. This means companies can be economically incentivized to pursue AGI development in scenarios where the expected outcome is net-negative for humanity.

What Are the Real Risks Beyond Job Loss?

While public discussion often focuses on unemployment and economic disruption, AI safety experts identify more severe existential concerns. Benjamin Todd, a leading AI risk researcher, explained that the conversation around AI risks has shifted significantly.

"We do think that if we have fully autonomous AI, loss of control of that AI is still one of our top-ranked risks. But we've added some extras to the list," Todd stated.

Benjamin Todd, AI Risk Researcher and 80,000 Hours

The expanded list of critical risks includes:

• Loss of Control: Fully autonomous AI systems that operate beyond human oversight or ability to shut down
• AI-Enabled Bioweapons: Advanced AI used to design or deploy biological weapons at scale
• Concentration of Power: AI systems enabling 24/7 surveillance of entire populations, synthesized into actionable intelligence for authoritarian control
• Gradual Disempowerment: Even without catastrophic failure, society could evolve in troubling ways as humans become increasingly dependent on AI decision-making

Todd emphasized that these risks differ fundamentally from near-term economic disruption. "Our focus is on risks that could be truly existential, permanent loss of civilization's potential," he noted.

Todd

Why Is the Timeline Suddenly Accelerating?

Google DeepMind CEO Demis Hassabis recently shifted the public conversation by articulating an unusually near-term timeline for AGI. Speaking at Google I/O 2026, Hassabis stated that humanity is standing in the "foothills of the singularity" and has only a few years left to prepare.

Hassabis

"My prediction that AGI could arrive in four years, or even sooner, reflects growing confidence that the industry has found the right technical path," Hassabis said.

Demis Hassabis, CEO of Google DeepMind

Hassabis broadened his earlier estimate, saying he still expects AGI around 2030, though he now sees 2029 as a possibility. His reasoning centers on the rapid advancement of AI agents, which he views as a "practice run" for more powerful systems. "You can imagine the agentic era in this next year is a little bit like a practice run," he explained.

Notably, Hassabis chose his language deliberately to provoke urgency among policymakers. "This is partly why I use some of the terms I used, yeah, which were a little bit provocative," he stated, adding that he is discussing possible safety measures with leaders at other top AI labs.

What Policy Changes Could Reduce Risk?

The University of Chicago research identifies several policy levers that could improve outcomes by changing the equilibrium incentives firms face. These approaches work by altering the market structure itself, not just by appealing to corporate responsibility.

• Consolidation: Reducing the number of competing firms can align incentives toward safety by eliminating the "race to the bottom" dynamic where fragmentation forces speed-focused decisions
• Resource Regulation: Controlling the total resources available for AI development can shift the trade-off between speed and safety in favor of safety
• Commitment Devices: Binding agreements among labs to coordinate on safety standards and development timelines, reducing the incentive to defect for competitive advantage
• Cautious Public Entry: Government involvement in AI development with explicit safety mandates can set norms and reduce pressure on private firms to cut safety corners

Hassabis expressed cautious optimism about government action, noting that "the federal government's tentative steps toward reprioritizing safety are a step in the right direction," referring to potential AI executive orders that would mandate testing before new models are released.

Hassabis

Why Should Economists Care About This Now?

Hassabis expressed frustration that the economic community has not fully engaged with the scale of potential transformation. "My economist friends, I feel, are still not taking this seriously enough," he remarked, emphasizing that the societal impact of AGI would rival the industrial revolution in scope.

Hassabis

The convergence of economic research and technical timelines suggests a narrow window for policy intervention. If AGI arrives within two to five years, as some researchers now predict, the decisions made today about market structure, resource allocation, and safety commitments will largely determine whether the transition benefits humanity or concentrates power in ways that are difficult to reverse.

The core insight from recent research is sobering: the problem is not just technical, and it cannot be solved by individual companies acting in isolation. Market incentives themselves must be reshaped to align private profit with public safety. Without structural change, competition alone will continue to push the industry toward speed, regardless of the risks involved.