Using AI to enhance societal decision making (article by Zershaaneh Qureshi)

Century of progress in a decade

Advanced AI may accelerate innovation so dramatically that decisions requiring years of deliberation must be made in months, exponentially increasing the risk of catastrophic missteps.

Pattern of institutional failure

History shows repeated failures to act on clear warnings, from climate change to COVID-19, suggesting current decision-making architectures are inadequate for higher-stakes futures.

Differential technology opportunity

We can intentionally accelerate safety-promoting AI tools—like verification and forecasting systems—so they arrive before dangerous capabilities, rather than waiting for default market development.

Epistemic enhancement

AI fact-checkers and forecasting systems could overcome human limitations in processing information and predicting outcomes, while moral reasoning tools might help society navigate complex ethical disagreements.

Coordination mechanisms

AI negotiation tools could simulate thousands of bargaining scenarios to find mutually beneficial agreements, while verification systems and structured transparency could enable trust between competing actors without total surveillance.

Near-term feasibility

Many of these applications may be buildable with current technology, representing low-hanging fruit compared to the massive investment in general AI capabilities.

Market gap argument

While some decision-making tools will emerge commercially, specific high-impact applications—like ethical reasoning aids or non-financial forecasting—lack market incentives and may arrive too late if pursued by default.

Managing acceleration risks

Targeting lower-risk applications like fact-checking rather than strategic planning minimizes the chance of advancing dangerous capabilities, and the benefits of better coordination likely outweigh small contributions to overall AI hype.

Democratizing access

To prevent power concentration, these tools must be widely distributed to institutions and the public, ensuring no single actor gains dangerous unilateral advantages through superior decision-making technology.