Many attendees predicted significant negative economic impacts from AI systems in the long term. 

We found that many attendees held a view of AI futures that was highly impactful, unevenly distributed, and focused on negative to neutral societal predictions for humans. Major economic themes included significant unemployment and labor displacement in specific sectors, widespread increases in wealth inequality, and dropping labor share. Other negative themes focused on a loss of purpose for humans, AI control of political systems, and negative impacts around health and safety (e.g. biochemical threats or LAWs).

In comparison, relatively little dialogue was concentrated on positive economic impacts, such as increasing productivity growth or widespread prosperity. Much of the economic causal modeling was neutral in theme - identifying potential variables and describing causal relationships rather than predicting societally beneficial or detrimental outcomes.

We suspect that this was influenced by our attendee base - we drew attendees roughly equally from established economists in traditional organizations and AI safety researchers. Within the AI safety community, there is a focus on identifying and mitigating risks from AI systems. As a result, the overall focus of the conference tended towards describing societal risks that could be improved, rather than describing prosperous outcomes. Attending economists tended to have more neutral and uncertain perspectives.

Simultaneously, attendees predicted limited global AI diffusion by 2030, indicating that they expect significant infrastructural and implementation barriers. 

Attendees found that despite far-reaching societal and economic expectations for AI technology during their worldbuilding exercises, their quantitative predictions for major global economic variables remained largely in line with current projections. By 2030, attendees predicted only a slight increase (6%) in physicians, a slight decrease (9%) in programmers, a moderate (2%) increase in global unemployment, a stable global labor share of GDP, and an increase in global median income ($2.56) largely commensurate with current trends. Even despite viewing AI as a revolutionary economic factor, our economists and AI researchers believe in the inertia and resilience of the global economy over short timescales.

Going into more detail, attendees suggested that even with rapid AI capability advancement, practical deployment will face significant implementation challenges that will constrain global AI diffusion through 2030. Physical infrastructure requirements like power sources, quantity of AI chips, and the availability of robotics will restrict diffusion globally, whereas reliable electricity, internet connectivity, and education will remain major limiting factors in developing regions. Systemic factors such as human trust in AI systems, institutional inertia, speed of retooling, and economic incentives will all negatively impact AI diffusion. 

We were surprised by the significant discrepancy between the worldbuilding results, which tended to describe dramatic and rapid global impacts, and the conservative nature of the group forecasts. One explanation is that attendees did not consider the 2030 timeframe to be a strict constraint, and preferred to describe their more long-term worldviews rather than focus purely on a 5-year timeframe. Another explanation is that the development of revolutionary technologies (e.g. the Internet) may cause widespread social and economic changes to occur, but still have relatively little impact on global economic indicators in the medium term. 

Economic impacts will be highly uneven by sector and region.

Results suggest that while AI will drive substantial economic changes by 2030, these impacts will be highly sector-specific and uneven. In sectors like transportation, tech, and creative industries, AI is expected to rapidly transform workflows and displace significant human labor. However, fields requiring physical skills (like surgery), high-prestige professional roles, and sectors with strong unions may see more AI augmentation than replacement, or even increased labor participation.

This transformation will be particularly uneven geographically, with AI-leading countries gaining significant advantages. AI advancement is expected to concentrate wealth among a small set of capital owners and tech companies, while increasing financial pressure on others through automation-driven unemployment. Developing countries will not merely be passive recipients of AI technologies, but could play key roles in the global AI supply chain by providing essential inputs such as natural resources or data. However, they will simultaneously face disruption in traditional service sectors like outsourcing and services..

There is a significant need for new economic metrics.

Our forecasting and economic modeling exercises revealed significant gaps in our ability to measure and track AI's economic impacts. The volume and diversity of forecasting questions generated by attendees - over 80 - underscores the complexity of this task. Attendees emphasized that conventional measures like GDP or labor share may fail to capture the full scope of AI-driven economic transformation, particularly given the potential divergence between an emerging "AI economy" and the traditional human economy. The OECD Better Life Index is an example of a more comprehensive approach, but attendees suggested even this needs adaptation to capture AI's impacts across varied dimensions such as physical safety, mental health, or economic stability.

Our conference had a strong emphasis on developing observable metrics that could serve as early indicators of AI's impacts. These include measures of AI system reliability, such as "the length of task chains before derailing," metrics of human-AI interaction like "tested variability of humans when using AI productively," measures of AI productivity, and broader societal indicators such as measures of social mobility or trust in political systems. The development of such metrics would be crucial for tracking AI's impact on both economic productivity and broader social outcomes.

There is a glaring lack of research on concrete AI economic policies.

A notable gap emerged in the discussions around policy interventions to address AI's economic impacts. While attendees discussed significant existing research describing potential economic impacts from AI systems, there was limited research referenced or discussion regarding concrete policy ideas to address these impacts. Conducting a separate literature review, our report authors have similarly found a dearth of practical, realistic economic policy proposals to mitigate the negative externalities of a post-AI economy.

In the previous section, we summarize some promising research questions on AI economic policies. Topics of neglected research include tax & revenue policy (corporate profit taxation, robot taxes, tax base distortions), social safety nets (UBI, job transition funds, AI reskilling, negative income taxes), predistributive policies & market incentives (public benefit corporations, insurance policies, procurement policies, due diligence), and public / private coordination (AI as a public utility, government managements / integrations).

There is a significant opportunity for expert consensus on the roles & responsibilities of government in a post-AI economy.

In general, there is a lack of urgent governmental awareness or concern around the broader economic impacts of advanced AI systems. Where there is awareness, there is still a gap in direction: policymakers, by and large, have no consensus or strong opinions about the best way to mitigate these upcoming concerns. Such economic policies are still outside the Overton window, largely due to uncertainties about future scenarios.

Despite this uncertainty, there is an opportunity today for leading economists, AI experts, and international policy researchers to identify and consolidate around a clear set of roles and responsibilities for governments in a post-AI economy. Such expert consensus would massively streamline the direction of governmental policy responses when significant AI economic impacts are inevitably realized.

To reach such a consensus, greater coordination is needed between various groups of AI economics research: traditional economists, AI safety & governance researchers, and governmental policymakers. Field-building in the form of conferences, collaborative consortiums, and academic journals will be critical. Frameworks and improved metrics to model and predict economic impacts must be developed. Significantly more research on concrete economic policies is needed.

Alignment and clarity on effective economic responses to upcoming AI economic impacts could improve the lives of billions of humans. These challenges demand immediate action - developing frameworks and economic strategies to ensure AI benefits humanity as a whole is among the most critical tasks facing society today.