The Future of Artificial Intelligence Policy Is the Future of Competing Demands

Overall, RAND researchers found tremendous potential upsides to the adoption of AI, along with an urgent need to manage downside risks. Looking at macroeconomic impacts, they saw potential gains for productivity growth and the federal debt but acknowledged that, even with the best analysis, there is limited visibility into AI's impacts on the labor market. Their analysis of early evidence suggests that AI is complementing labor productivity, but there is a risk of future labor displacement as capabilities advance. In their analysis of the macroeconomic implications of AI, they found that, by 2035, moderate AI-driven productivity gains could boost real per-capita gross domestic product by nearly $7,000.

But these productivity gains must be balanced against questions about how human flourishing is being supported and what that means for the social contract. Risks and challenges from AI adoption that extend to privacy, oversight, and potential societal polarization require careful management and the protection of individual rights and democratic processes. Managing the challenges of AI adoption while enabling its opportunities will also require new metrics for understanding the complex, multidimensional impacts of AI adoption on the labor market, macroeconomy, and society.

Read More: “Macroeconomic Implications of Artificial Intelligence” Read More: “Artificial Intelligence and the Social Contract”

The financial services sector stands as a pioneer in AI adoption. For example, AI applications in risk management and fraud detection have already shown promise helping to identify anomalies in transaction data that might indicate fraud and enabling financial institutions to automatically check compliance and provide alerts for potential breaches. In the future, breakthrough AI capabilities could enable instant lending decisions and predictive trading.

But these advances could introduce novel forms of systemic risk and potentially hinder oversight and accountability. Additionally, as financial institutions deploy increasingly sophisticated AI systems, they will need access to vast amounts of sensitive customer information to function effectively. Furthermore, the nascent integration of AI and the diverse landscape of cryptocurrencies and digital assets could disrupt existing financial models, particularly through disintermediation, promoting a need for adaptive regulatory frameworks to manage systemic risks.

Read More: “AI Capabilities and Uptake: Impacts on Healthcare, Financial Services, Climate, and Transportation” Read More: “Artificial Intelligence and Crypto in Financial Services”

In the health care industry, AI advances have demonstrated enormous potential in such areas as drug development. In health care, the AI-based AlphaFold model developed by Google DeepMind represented a breakthrough in solving the problem of understanding protein and gene structures, and its creators were awarded the Nobel Prize in chemistry. Although no AI-designed pharmaceuticals have yet been shown effective and safe in clinical trials, dozens are in the development pipeline, suggesting that AI could become standard in drug discovery. In addition, existing AI algorithms that are used for the early detection of cancer, heart failure, or sepsis have outperformed humans; however, the adoption of these capabilities has been limited. Growing uptake and trust in AI technologies could enable more early interventions and cost savings, but it is uncertain whether such a shift would help physicians by reducing tasks or cause burnout because of liability and verification responsibilities. There is also limited evidence regarding the safety and effectiveness of many AI applications in health care. Very few AI-based tools are evaluated in randomized trials that measure health impacts, which is the gold standard for both doctors and the public.

AI has the potential to streamline decarbonization and transition planning and improve the efficiency of existing energy systems. AI also has the potential to help accelerate fossil fuel exploration and extraction. This overarching trade-off has coincided with a slower rate of AI adoption in the climate and energy space than in other sectors. Nevertheless, AI is helping consumers improve energy efficiency and lower their costs while allowing energy firms to forecast demand and dynamically source energy supplies based on demand fluctuations. Additionally, researchers are developing tools to enable autonomous grid management. AI is also advancing climate models and weather forecasts to support more-robust climate risk assessments, which raises questions about how financial institutions and insurance and real estate companies will respond. And across every sector, AI is already leading to massive energy demands; these needs have the potential to compete with community and residential needs, leading to higher prices and even energy shortages. At the same time, the need to meet new energy demands could result in higher rates of research, investment, and development that encourage low-carbon energy innovations.

For these sectors and many others, concerns about widespread job displacement, significant invasion of data privacy, cybersecurity failures, bias and increasing social inequity, and misalignment with human values will be a defining feature of the emerging AI landscape. But it is not just individual sector impacts that policymakers will need to confront. Cascading failures across interconnected systems could amplify AI risks because disruptions in one sector can propagate widespread impacts through digital, physical, and social networks. Furthermore, traditional economic metrics (such as gross domestic product) are inadequate for measuring AI’s multidimensional impacts on human well-being, social inequalities, and environmental sustainability.

Read More: “AI Capabilities and Uptake: Impacts on Healthcare, Financial Services, Climate, and Transportation” Read More: “Rethinking Social and Economic Policy in the Age of General-Purpose Artificial Intelligence”