Appendix A — Mondo International

In recent years, the intersection of artificial intelligence (AI) and economics has become an area of growing interest, with AI being increasingly employed to enhance the capabilities of traditional economic models. AI offers a multi-faceted approach, drawing on three main families—explanatory, predictive, and generative AI—each contributing uniquely to the refinement and expansion of economic modeling. These families of AI, when integrated with traditional econometric models such as the Phillips Curve, enable economists to leverage vast amounts of data and sophisticated algorithms to improve the accuracy, scalability, and scope of economic forecasts. Explanatory AI plays a crucial role in enhancing the interpretability of economic models by offering insights into the causal relationships between variables. Traditional econometric models, often limited by assumptions of linearity or predefined interactions, benefit from explanatory AI’s ability to process a broader array of inputs, including both structured and unstructured data. This type of AI enables economists to extract deeper insights from complex datasets, making it possible to explain not only the outcomes but also the factors driving those outcomes. For instance, explanatory AI can provide a more nuanced understanding of inflation by examining the interplay between various macroeconomic indicators, consumer sentiment, and international trade flows. This enhances the explanatory power of models like the Phillips Curve, allowing economists to develop more sophisticated narratives that account for real-world complexities. In economic environments characterized by uncertainty or volatility, explanatory AI offers a tool for untangling intricate dynamics that traditional models may overlook.

Predictive AI, another key branch, focuses on improving the accuracy and reliability of economic forecasts. The traditional reliance on historical data and static assumptions in economic models limits their predictive capabilities, especially in fast-evolving contexts. Predictive AI, however, excels in processing vast datasets, including real-time high-frequency indicators such as stock market movements, social media trends, and global economic news. This enhances the predictive power of models, particularly in long-term economic forecasts such as inflation or GDP growth. A study conducted by the Federal Reserve Bank of St. Louis demonstrated that AI models, when trained on diverse datasets, outperform human forecasters in predicting macroeconomic trends across multiple time horizons (Jean, Bartlett, & Turcotte, 2024). These AI-driven predictions are particularly valuable when applied across multiple geographies, offering insights at both national and international levels. As AI evolves, predictive models are likely to become even more precise, incorporating an ever-expanding array of data sources to better account for the complexity of global economic systems.

Generative AI extends the potential of AI in economics by automating the generation of economic reports and analyses. Traditionally, generating forecasts and corresponding policy reports for multiple countries or sectors would require significant time and resources. Generative AI automates this process, scaling the application of models like the Phillips Curve across dozens or even hundreds of countries in a fraction of the time. For instance, the generation of inflation forecasts for 193 countries, which would otherwise necessitate a team of economists, can be accomplished almost instantaneously using generative AI. And it goes beyond what people are now familiar with, we can indeed use Retrieval-Augmented Generation (RAG) models to tap into our specific data. This level of scalability is particularly useful for global institutions such as the International Monetary Fund (IMF) or the World Bank, where consistent and comparative analyses across multiple regions are essential for policy recommendations. Generative AI can also simulate hypothetical economic scenarios, allowing policymakers to test different policy interventions or economic conditions before making decisions. While generative AI provides remarkable efficiency, it is important to recognize its limitations, especially in cases where human expertise is required to interpret the results and provide context-specific recommendations. As noted by the Desjardins economic research team, AI models still rely on historical data and may inadvertently “cheat” by using future information when generating forecasts, necessitating human oversight to ensure accuracy and reliability (Desjardins, 2024).

The integration of these three AI families—explanatory, predictive, and generative—into economic modeling offers significant advantages in terms of scalability. Traditional econometric models often require extensive calibration for individual countries or regions, making large-scale, cross-country comparisons difficult and time-consuming. AI, by contrast, enables the same models to be scaled with minimal adjustment across multiple geographies. This capability is especially valuable for institutions conducting cross-country comparative studies, allowing for the production of consistent and comparable results across different regions. As AI continues to advance, its capacity to incorporate new data sources, refine model accuracy, and provide tailored insights at scale will make it an indispensable tool for economic forecasting and analysis.

Despite the many benefits of incorporating AI into economic modeling, challenges remain. One significant challenge is the “black-box” nature of many AI models, where the decision-making processes are opaque and difficult to interpret. In economics, where transparency and theoretical grounding are critical, this can pose a problem. Economists must be able to explain the rationale behind their models and predictions, particularly when these are used to inform policy decisions. Therefore, the development of explainable AI models, which make the relationships between variables and the underlying mechanics of predictions more transparent, is a crucial area of ongoing research. Another challenge lies in the quality of the data used to train AI models. Predictive and generative AI models are only as good as the data they are trained on, and biases or gaps in the data can lead to skewed predictions or flawed conclusions. This is particularly relevant in global contexts where the quality and availability of economic data vary widely between countries.

AI offers significant potential to enhance traditional economic models by leveraging explanatory, predictive, and generative capabilities. These three families of AI work together to improve the accuracy, scalability, and interpretability of economic forecasts. While explanatory AI helps unravel the complex relationships between economic variables, predictive AI enhances the precision of forecasts, and generative AI allows for large-scale, automated analyses across multiple geographies. However, as with any powerful tool, AI must be used with caution. Over-reliance on AI-generated forecasts without human oversight can lead to unforeseen errors or biases, particularly in contexts where data quality is uneven. Therefore, AI should be viewed as a complementary tool that enhances, rather than replaces, the expertise of economists. As AI technologies continue to evolve, their role in economics will likely expand, offering new possibilities for refining both economic theory and policy.

Enhancing Economic Models with AI

The use of artificial intelligence in economics is increasingly recognized as a valuable tool for enhancing the accuracy and scope of traditional economic models. One of the primary ways AI is employed in this field is by integrating it with existing models to improve predictive accuracy. This approach allows economists to retain the core theoretical frameworks of widely accepted models, such as the Phillips Curve or Keynesian models, while enhancing their forecasting abilities through the processing of vast amounts of data. AI can, in this sense, serve as a complementary tool that makes existing models more efficient and scalable, especially in contexts that involve large datasets and complex relationships between economic variables. Economic forecasting has traditionally relied on econometric models built around established theories, using a limited number of variables and based on assumptions that can often be simplified due to the limitations of the models themselves. For instance, when applied to macroeconomic indicators such as inflation or unemployment, traditional models typically rely on historical data and predefined relationships between variables. These models are powerful but constrained by their linear nature and their dependency on specific assumptions. The integration of AI allows for more complex, non-linear relationships to be modeled, providing a richer understanding of the underlying dynamics. AI can incorporate real-time data, social media sentiment, and global economic news, offering more nuanced predictions that account for rapidly changing economic environments. A particularly notable application of AI is its ability to scale economic models across multiple geographies or scenarios. For example, applying the Phillips Curve to inflation predictions for 193 countries would require a tremendous amount of time and resources if done manually by economists. AI, however, can automate this process, generating predictions and insights for all these countries at once, thus allowing economists to conduct large-scale comparative studies in a fraction of the time traditionally required. Such scalability is crucial in a globalized world where economic interdependence means that understanding the macroeconomic environment in one country often requires considering the contexts of many others. Furthermore, as AI models are trained on broader datasets, their ability to predict outcomes in different contexts increases, making them invaluable for international institutions like the World Bank or International Monetary Fund (IMF) that require cross-country economic analyses. Generative AI takes these capabilities a step further by automating not just the analysis but also the generation of reports and interpretations based on model outputs. This capacity could revolutionize the way economic analyses are conducted and presented. Generative AI can automatically generate country-specific reports or forecast analyses that economists can then review, saving them valuable time. This process would have particular utility for producing annual reports or forecasting documents for policymakers, where consistency across multiple reports is essential, yet achieving this through traditional means can be tedious and error-prone. AI can thus ensure that such reports maintain a high level of precision and consistency while allowing economists to focus on the higher-order task of interpreting the results and making policy recommendations. The potential of AI to improve economic predictions is supported by empirical evidence. A study conducted by the Federal Reserve Bank of St. Louis (Jean, Bartlett, & Turcotte, 2024) demonstrated that AI models, when applied to inflation forecasting, produced lower forecast errors than traditional models across a range of countries. The ability of AI to incorporate diverse data sources, including high-frequency economic indicators and even non-traditional data such as satellite images or social media posts, enhances the robustness of these forecasts. This is particularly important in an era of rapid globalization and technological change, where traditional economic models may struggle to keep up with the pace and complexity of economic developments. Despite these advantages, the integration of AI into economic models does come with certain challenges. One of the most significant concerns is the so-called “black box” problem, where the internal workings of AI models are not easily interpretable by human experts. In economics, where transparency and theoretical justification are key, this can be a serious limitation. Economists need to be able to explain why certain predictions are being made, especially when these predictions are used to inform policy decisions that can have wide-reaching effects. To address this issue, there is growing interest in developing explainable AI models that provide not only accurate predictions but also insights into the causal relationships and mechanisms driving these predictions. This would allow economists to use AI as a tool that not only improves predictive power but also contributes to a deeper understanding of economic phenomena. Another potential issue is that AI models are highly dependent on the quality and quantity of the data they are trained on. In cases where data is limited or biased, AI models can produce misleading or skewed predictions. This is particularly relevant in economic forecasting, where the availability of reliable data can vary significantly across countries or sectors. As a result, economists must exercise caution when using AI models in contexts where data quality is a concern, ensuring that the models are calibrated and validated appropriately to avoid erroneous conclusions. Furthermore, AI’s ability to “cheat” by using future data to generate historical predictions, as identified by researchers at Desjardins (2024), illustrates the need for rigorous oversight and refinement in how AI models are deployed in economic contexts. The integration of AI into existing economic models holds great promise for improving the accuracy, efficiency, and scalability of economic predictions. By enhancing traditional models with the processing power and flexibility of AI, economists can produce more nuanced forecasts that are better suited to the complexities of a globalized world. However, to fully realize the potential of AI in economics, researchers and policymakers must continue to address challenges related to model transparency, data quality, and ethical considerations. As AI technology continues to evolve, its role in economic forecasting and analysis is likely to grow, making it an indispensable tool for both academics and practitioners.

Challenging Economic Theories with Inductive Approaches

The second avenue for the use of artificial intelligence in economics involves challenging traditional economic theories through inductive approaches. Inductive reasoning, which involves drawing general conclusions from specific data, is not new to economics. However, AI has the potential to elevate this method by processing vast amounts of data and identifying patterns that were previously undetectable through traditional statistical methods. This section explores how AI can be used to challenge and reshape economic theories, with a particular emphasis on its capacity to generate new hypotheses and test existing models. Inductive approaches in economics have historically been used to develop new theoretical frameworks based on empirical observations. A well-known example is the gravity model of trade, which emerged from the observation that trade between two countries is positively correlated with their economic size and negatively correlated with the distance between them. This model was not initially derived from a specific theory but from empirical regularities in the data. Similarly, AI can be employed to identify patterns and correlations in economic data that may challenge or complement existing theories. For instance, machine learning algorithms can process large datasets encompassing various economic indicators—ranging from consumer behavior to international trade flows—and reveal relationships that may not fit within the confines of traditional economic models. One of the key advantages of using AI for inductive reasoning in economics is its capacity for handling vast and complex datasets. Economic data is increasingly diverse, encompassing not only quantitative indicators like GDP or inflation rates but also qualitative data such as sentiment analysis from social media or policy documents. Traditional econometric models are often limited in their ability to incorporate these diverse data sources, as they rely on linear relationships and predefined assumptions. AI, by contrast, is capable of processing non-linear relationships and unstructured data, thus offering a more flexible framework for exploring new theoretical possibilities. This flexibility allows AI to identify new patterns in the data that may not conform to established economic theories, thereby prompting economists to rethink their assumptions and develop new hypotheses. Moreover, AI’s capacity for inductive reasoning enables economists to test existing theories in new ways. Traditionally, economic theories are tested deductively, with models being built based on theoretical assumptions and then tested against empirical data. However, AI allows for a more data-driven approach, where models are generated inductively based on patterns observed in the data, and then tested to see how well they fit with existing theories. This approach is particularly useful for exploring areas of economics where traditional models may fall short, such as the impact of technological innovation on labor markets or the dynamics of global supply chains in the context of geopolitical instability. A specific area where AI-driven inductive reasoning could have a significant impact is in macroeconomic modeling. Macroeconomic models, such as those used by central banks to forecast inflation or unemployment, are often based on a set of assumptions about how different variables interact. However, these models can be limited in their ability to account for the complexity of modern economies, where factors such as global trade dynamics, technological change, and political risk play an increasingly important role. AI offers the potential to incorporate these factors into macroeconomic models in new ways, by identifying patterns in the data that may not have been captured by traditional models. This could lead to the development of new macroeconomic theories that better reflect the realities of a rapidly changing global economy. While the potential for AI to challenge and reshape economic theories is significant, it is important to recognize the limitations of this approach. One key limitation is the “black-box” nature of many AI algorithms, which makes it difficult to understand how specific conclusions are reached. In economics, where the ability to explain and justify theoretical assumptions is critical, this lack of transparency can be a significant drawback. Economists must therefore be cautious in adopting AI-driven inductive approaches, ensuring that any new hypotheses or models generated by AI are both interpretable and theoretically sound. Furthermore, AI-driven inductive reasoning is only as good as the data on which it is based. If the data used to train AI models is biased or incomplete, the conclusions generated by the model may also be flawed. This is particularly relevant in the context of global economic data, where disparities in data availability and quality between countries can lead to skewed results. For example, countries with more comprehensive economic data may disproportionately influence the conclusions drawn by AI models, leading to theoretical frameworks that do not fully account for the experiences of less data-rich economies. AI offers significant potential to challenge and reshape traditional economic theories through inductive approaches. By processing vast amounts of diverse data and identifying new patterns, AI can prompt economists to rethink existing models and develop new hypotheses. However, the use of AI in this context must be approached with caution, given the limitations of AI algorithms and the potential for bias in the data. Future research should focus on refining AI-driven inductive approaches in economics, ensuring that the insights generated are both accurate and theoretically sound.

Case Study: The Mondo Project at CIRANO

The Mondo International project exemplifies a comprehensive approach to integrating artificial intelligence in economics by employing the three main families of AI: explanatory, predictive, and generative, including RAG approaches. Each of these AI categories contributes uniquely to the analysis of economic data, allowing for a more multifaceted understanding of global economic trends and decision-making. Explanatory AI, which focuses on understanding and interpreting data, is used extensively in the Mondo International project to provide insights into the causal relationships between economic variables. This branch of AI is critical for uncovering the underlying factors that drive economic outcomes, allowing economists and policymakers to move beyond mere correlations and delve into the mechanisms that explain observed patterns. In practice, explanatory AI is applied at all levels of analysis—macro, meso, and micro—helping to clarify why certain economic phenomena occur. For instance, by analyzing sectoral and regional data, explanatory AI can reveal why specific industries perform differently in response to similar economic policies or external shocks. This type of AI assists in providing clearer explanations for the complex, interconnected dynamics of global supply chains, consumer behavior, and firm-level productivity. Predictive AI, on the other hand, is central to the project’s ability to forecast future economic trends. By employing machine learning models trained on extensive datasets, predictive AI enables Mondo International to generate highly accurate forecasts for key macroeconomic indicators such as GDP growth, inflation, and unemployment. These models are not limited to traditional economic variables but incorporate real-time, high-frequency data from diverse sources, such as social media sentiment, market trends, and global news. Predictive AI allows for a more dynamic and responsive approach to economic forecasting, making it especially valuable in volatile economic environments where quick adjustments are necessary. In the Mondo International project, predictive AI models are applied not only to national economies but also to individual sectors and firms, allowing for granular, actionable forecasts that can inform both policymakers and business leaders. Generative AI represents the most innovative application of artificial intelligence within the Mondo International project. It automates the generation of reports, summaries, and even hypothetical scenarios based on the model outputs, allowing for the efficient dissemination of insights across a wide range of contexts. For example, generative AI can produce country-specific economic analyses based on the same theoretical models, providing tailored insights for different regions without requiring significant manual effort. This scalability makes generative AI an invaluable tool for organizations that need to conduct large-scale comparative analyses across multiple countries or industries. Additionally, generative AI can simulate hypothetical economic scenarios—such as the impact of policy changes or external shocks—enabling economists and policymakers to explore potential outcomes and make more informed decisions. By automating the generation of these analyses, generative AI enhances the efficiency and reach of the Mondo International project, allowing it to provide real-time insights to a global audience. The combination of these three AI families—explanatory, predictive, and generative—gives the Mondo International project a unique edge in economic analysis. Explanatory AI provides the interpretive framework necessary to understand complex economic relationships, predictive AI offers highly accurate forecasts, and generative AI automates the production and dissemination of insights. Together, these AI capabilities allow the project to tackle the challenges of modern economic analysis in a comprehensive and scalable manner. Whether analyzing macroeconomic trends, sectoral dynamics, or firm-level behavior, the Mondo International project demonstrates how AI can transform the way economists approach data, forecast outcomes, and communicate insights. The Mondo International project’s use of the three families of AI—explanatory, predictive, and generative—illustrates the profound potential of AI to reshape economic analysis. By combining these different AI techniques, the project is able to provide deeper, more accurate insights into global economic trends, while also offering scalable solutions for real-time policy analysis and decision-making. As AI technologies continue to evolve, their application in economics, as demonstrated by Mondo International, will play an increasingly central role in shaping both economic theory and practice. The primary goal of the Mondo International project is to leverage AI to address some of the long-standing challenges in economic data analysis. Economic data is notoriously complex, with different levels of aggregation and often missing, inconsistent, or outdated data points. Traditional methods for handling such data are often time-consuming and prone to error. Mondo International seeks to address these issues by employing advanced AI techniques to clean, integrate, and analyze data across multiple levels. In doing so, the project provides a more coherent and comprehensive view of global economic trends, offering insights that might otherwise be overlooked. One of the key features of the Mondo International project is its ability to analyze data at multiple levels of economic activity. At the macro level, AI is used to analyze national economic indicators such as GDP, inflation, and unemployment rates. These indicators are typically analyzed over time and across countries, allowing for the identification of global economic trends and the development of more accurate forecasting models. For example, the use of machine learning algorithms in this context allows the project to generate predictions for GDP growth across different countries, incorporating not only traditional economic indicators but also non-traditional data sources such as social media sentiment or news reports. This multidimensional analysis enables policymakers to anticipate economic shifts with greater precision. At the meso level, Mondo International focuses on sectoral data, examining how different industries or regions respond to economic policies or external shocks. This level of analysis is particularly useful for understanding the uneven effects of economic change across different parts of the economy. For instance, AI-driven analysis can reveal how a policy aimed at stimulating growth in one sector may inadvertently cause a slowdown in another. By identifying such interactions, Mondo International provides valuable insights for policymakers, allowing for more targeted and effective economic interventions. Finally, at the micro level, the project uses AI to analyze firm-level data, exploring how individual companies respond to economic conditions and policy changes. This level of analysis is particularly important for understanding the drivers of innovation and productivity within firms. AI can identify patterns in the behavior of firms that may not be immediately obvious, such as the relationship between investment in technology and long-term profitability. Additionally, by analyzing vast amounts of firm-level data, AI can help policymakers and business leaders make more informed decisions about where to allocate resources or how to support innovation. The methodologies employed by Mondo International are diverse, reflecting the complexity of the economic data being analyzed. Machine learning algorithms are used to identify patterns in the data and generate predictive models. These algorithms are trained on vast datasets, allowing them to learn from past trends and make more accurate forecasts about future economic conditions. Natural language processing (NLP) techniques are also used to analyze unstructured data, such as news articles or social media posts, which provide valuable context for understanding economic trends. For example, NLP can be used to gauge public sentiment about a particular economic policy, providing real-time feedback that can be integrated into economic models. One of the most significant findings of the Mondo International project is the identification of hidden correlations between different economic variables. By analyzing data from multiple sources and at different levels, AI has uncovered relationships that were previously unrecognized by traditional economic models. For instance, the project has identified significant links between consumer behavior in one region and industrial output in another, suggesting that global supply chains are more interconnected than previously thought. These findings have important implications for economic theory, particularly in areas such as trade and globalization, where traditional models often assume relatively simple relationships between economic variables. The implications of the Mondo International project for economics are profound. By employing AI to analyze economic data in new ways, the project is pushing the boundaries of what is possible in economic forecasting and policy analysis. The ability to integrate data from multiple sources and analyze it at different levels provides a more nuanced and accurate picture of economic conditions, allowing for more informed decision-making. This has particular relevance in the context of global economic challenges, such as the COVID-19 pandemic, where traditional economic models have struggled to account for the complexity and unpredictability of the situation. AI, with its ability to process vast amounts of data and identify hidden patterns, offers a powerful tool for navigating such challenges. The Mondo International project represents a significant advance in the application of AI to economics. By integrating AI techniques with traditional economic analysis, the project provides deeper insights into macro, meso, and micro-level economic trends. The ability to analyze data at multiple levels and from multiple sources offers a more comprehensive understanding of the global economy, with important implications for both economic theory and policy. As AI continues to evolve, projects like Mondo International will play an increasingly important role in shaping the future of economic analysis and decision-making. One of the most recent advancements in generative AI that has significant implications for economic analysis is Retrieval-Augmented Generation (RAG). RAG combines the strengths of traditional information retrieval systems, such as databases, with the generative capabilities of large language models (LLMs). This framework allows AI to access and incorporate real-time data, enhancing the contextual relevance, factual accuracy, and up-to-date nature of the generated outputs. RAG operates through a two-step process. First, it retrieves relevant information from an external knowledge base, such as web pages, databases, or proprietary economic data sources. This retrieval is facilitated by powerful search algorithms that query vast datasets to find the most relevant information. Second, this retrieved information is integrated into the LLM’s generative process, enriching the AI’s responses by providing it with a more comprehensive understanding of the topic at hand. By doing so, RAG ensures that the generated economic analyses or forecasts are not only theoretically sound but also grounded in the most current and relevant data available. This capability is particularly beneficial in the context of economic forecasting, where timely and accurate data is crucial. Traditional models often rely on historical data, which may not always reflect the rapid changes in today’s global economy. RAG, by contrast, enables models to incorporate real-time, high-frequency economic indicators such as financial market data, consumer sentiment from social media, or breaking news reports, ensuring that forecasts remain responsive to the latest developments. For example, in the Mondo International project, RAG could be used to enhance the generation of country-specific economic reports. By retrieving real-time data on inflation trends, commodity prices, or trade flows from external sources, RAG ensures that the generated reports are not only consistent with historical models but also reflect the most up-to-date global conditions. This ability to integrate and process real-time information addresses one of the main challenges of traditional AI models—relying solely on pre-trained, static knowledge bases—and offers more dynamic, data-driven insights for decision-making. RAG also helps address the “hallucination” problem often associated with generative models, where AI might generate inaccurate or misleading information. By grounding the generated outputs in factual, retrieved data, RAG promotes factual consistency, making it a particularly valuable tool in applications where accuracy is paramount, such as economic forecasting, policymaking, and financial analysis. The integration of RAG into generative AI models represents a significant step forward in enhancing the precision and reliability of AI-driven economic insights. As RAG enables AI models to access and integrate real-time data from a variety of sources, it will likely become an indispensable tool for institutions like central banks, global organizations, and policymakers who require accurate, up-to-date economic analyses to navigate the complexities of the modern global economy.

Conclusion

In conclusion, the integration of artificial intelligence into economics represents a significant evolution in how economic models are developed, tested, and applied. AI, through its three families—explanatory, predictive, and generative—offers a range of tools that complement and extend traditional econometric models. Explanatory AI enhances the interpretability of complex relationships within data, predictive AI improves forecasting accuracy by processing vast and diverse data sources, and generative AI automates and scales the production of economic analyses, enabling more efficient cross-country comparisons and scenario generation. These capabilities have already begun to reshape the landscape of economic research and policy-making by providing new ways to understand and anticipate global economic trends. The Mondo International project stands as an interesting case study in this broader integration of AI within economics. By applying AI techniques across multiple levels of economic data—macro, meso, and micro—Mondo International demonstrates how AI can deliver deeper, more comprehensive insights that go beyond what traditional models can offer. Its use of all three AI families allows the project to analyze large-scale data while also producing granular, specific insights for policymakers and business leaders alike. The project’s ability to scale complex economic analyses across multiple countries highlights how AI can make global economic research more efficient and accessible, thus improving decision-making on an international scale. Beyond its immediate applications, the Mondo International project also raises important questions about the future of economic theory and practice. As AI continues to evolve, projects like Mondo will challenge established economic theories by uncovering new relationships within data and generating hypotheses that could reshape the field. This inductive approach to theory development has the potential to expand our understanding of economics in ways previously unimagined, forcing researchers and policymakers to rethink their assumptions and strategies. Overall, the intersection of AI and economics is not just a technical enhancement but a paradigm shift in how we approach economic data and decision-making. Projects like Mondo International offer a glimpse into a future where AI not only improves the precision of economic forecasts but also generates entirely new ways of thinking about the economy, offering a more dynamic and data-driven approach to understanding and managing global challenges.

References




Jean, J., Bartlett, R., & Turcotte, S. (2024). "Inflation Forecasting Using AI: A Comparative Analysis." Federal Reserve Bank of St. Louis.


Desjardins Economic Research (2024). "Point de vue économique." Desjardins, Études économiques. 
Turcotte, S. (2024). "Economic Forecasting and AI: A Comparative Study." HEC Montréal.

# Mondo International In recent years, the intersection of artificial intelligence (AI) and economics has become an area of growing interest, with AI being increasingly employed to enhance the capabilities of traditional economic models. AI offers a multi-faceted approach, drawing on three main families—explanatory, predictive, and generative AI—each contributing uniquely to the refinement and expansion of economic modeling. These families of AI, when integrated with traditional econometric models such as the Phillips Curve, enable economists to leverage vast amounts of data and sophisticated algorithms to improve the accuracy, scalability, and scope of economic forecasts. Explanatory AI plays a crucial role in enhancing the interpretability of economic models by offering insights into the causal relationships between variables. Traditional econometric models, often limited by assumptions of linearity or predefined interactions, benefit from explanatory AI’s ability to process a broader array of inputs, including both structured and unstructured data. This type of AI enables economists to extract deeper insights from complex datasets, making it possible to explain not only the outcomes but also the factors driving those outcomes. For instance, explanatory AI can provide a more nuanced understanding of inflation by examining the interplay between various macroeconomic indicators, consumer sentiment, and international trade flows. This enhances the explanatory power of models like the Phillips Curve, allowing economists to develop more sophisticated narratives that account for real-world complexities. In economic environments characterized by uncertainty or volatility, explanatory AI offers a tool for untangling intricate dynamics that traditional models may overlook. Predictive AI, another key branch, focuses on improving the accuracy and reliability of economic forecasts. The traditional reliance on historical data and static assumptions in economic models limits their predictive capabilities, especially in fast-evolving contexts. Predictive AI, however, excels in processing vast datasets, including real-time high-frequency indicators such as stock market movements, social media trends, and global economic news. This enhances the predictive power of models, particularly in long-term economic forecasts such as inflation or GDP growth. A study conducted by the Federal Reserve Bank of St. Louis demonstrated that AI models, when trained on diverse datasets, outperform human forecasters in predicting macroeconomic trends across multiple time horizons (Jean, Bartlett, & Turcotte, 2024). These AI-driven predictions are particularly valuable when applied across multiple geographies, offering insights at both national and international levels. As AI evolves, predictive models are likely to become even more precise, incorporating an ever-expanding array of data sources to better account for the complexity of global economic systems. Generative AI extends the potential of AI in economics by automating the generation of economic reports and analyses. Traditionally, generating forecasts and corresponding policy reports for multiple countries or sectors would require significant time and resources. Generative AI automates this process, scaling the application of models like the Phillips Curve across dozens or even hundreds of countries in a fraction of the time. For instance, the generation of inflation forecasts for 193 countries, which would otherwise necessitate a team of economists, can be accomplished almost instantaneously using generative AI. And it goes beyond what people are now familiar with, we can indeed use Retrieval-Augmented Generation (RAG) models to tap into our specific data. This level of scalability is particularly useful for global institutions such as the International Monetary Fund (IMF) or the World Bank, where consistent and comparative analyses across multiple regions are essential for policy recommendations. Generative AI can also simulate hypothetical economic scenarios, allowing policymakers to test different policy interventions or economic conditions before making decisions. While generative AI provides remarkable efficiency, it is important to recognize its limitations, especially in cases where human expertise is required to interpret the results and provide context-specific recommendations. As noted by the Desjardins economic research team, AI models still rely on historical data and may inadvertently "cheat" by using future information when generating forecasts, necessitating human oversight to ensure accuracy and reliability (Desjardins, 2024). The integration of these three AI families—explanatory, predictive, and generative—into economic modeling offers significant advantages in terms of scalability. Traditional econometric models often require extensive calibration for individual countries or regions, making large-scale, cross-country comparisons difficult and time-consuming. AI, by contrast, enables the same models to be scaled with minimal adjustment across multiple geographies. This capability is especially valuable for institutions conducting cross-country comparative studies, allowing for the production of consistent and comparable results across different regions. As AI continues to advance, its capacity to incorporate new data sources, refine model accuracy, and provide tailored insights at scale will make it an indispensable tool for economic forecasting and analysis. Despite the many benefits of incorporating AI into economic modeling, challenges remain. One significant challenge is the "black-box" nature of many AI models, where the decision-making processes are opaque and difficult to interpret. In economics, where transparency and theoretical grounding are critical, this can pose a problem. Economists must be able to explain the rationale behind their models and predictions, particularly when these are used to inform policy decisions. Therefore, the development of explainable AI models, which make the relationships between variables and the underlying mechanics of predictions more transparent, is a crucial area of ongoing research. Another challenge lies in the quality of the data used to train AI models. Predictive and generative AI models are only as good as the data they are trained on, and biases or gaps in the data can lead to skewed predictions or flawed conclusions. This is particularly relevant in global contexts where the quality and availability of economic data vary widely between countries. AI offers significant potential to enhance traditional economic models by leveraging explanatory, predictive, and generative capabilities. These three families of AI work together to improve the accuracy, scalability, and interpretability of economic forecasts. While explanatory AI helps unravel the complex relationships between economic variables, predictive AI enhances the precision of forecasts, and generative AI allows for large-scale, automated analyses across multiple geographies. However, as with any powerful tool, AI must be used with caution. Over-reliance on AI-generated forecasts without human oversight can lead to unforeseen errors or biases, particularly in contexts where data quality is uneven. Therefore, AI should be viewed as a complementary tool that enhances, rather than replaces, the expertise of economists. As AI technologies continue to evolve, their role in economics will likely expand, offering new possibilities for refining both economic theory and policy. Enhancing Economic Models with AI The use of artificial intelligence in economics is increasingly recognized as a valuable tool for enhancing the accuracy and scope of traditional economic models. One of the primary ways AI is employed in this field is by integrating it with existing models to improve predictive accuracy. This approach allows economists to retain the core theoretical frameworks of widely accepted models, such as the Phillips Curve or Keynesian models, while enhancing their forecasting abilities through the processing of vast amounts of data. AI can, in this sense, serve as a complementary tool that makes existing models more efficient and scalable, especially in contexts that involve large datasets and complex relationships between economic variables. Economic forecasting has traditionally relied on econometric models built around established theories, using a limited number of variables and based on assumptions that can often be simplified due to the limitations of the models themselves. For instance, when applied to macroeconomic indicators such as inflation or unemployment, traditional models typically rely on historical data and predefined relationships between variables. These models are powerful but constrained by their linear nature and their dependency on specific assumptions. The integration of AI allows for more complex, non-linear relationships to be modeled, providing a richer understanding of the underlying dynamics. AI can incorporate real-time data, social media sentiment, and global economic news, offering more nuanced predictions that account for rapidly changing economic environments. A particularly notable application of AI is its ability to scale economic models across multiple geographies or scenarios. For example, applying the Phillips Curve to inflation predictions for 193 countries would require a tremendous amount of time and resources if done manually by economists. AI, however, can automate this process, generating predictions and insights for all these countries at once, thus allowing economists to conduct large-scale comparative studies in a fraction of the time traditionally required. Such scalability is crucial in a globalized world where economic interdependence means that understanding the macroeconomic environment in one country often requires considering the contexts of many others. Furthermore, as AI models are trained on broader datasets, their ability to predict outcomes in different contexts increases, making them invaluable for international institutions like the World Bank or International Monetary Fund (IMF) that require cross-country economic analyses. Generative AI takes these capabilities a step further by automating not just the analysis but also the generation of reports and interpretations based on model outputs. This capacity could revolutionize the way economic analyses are conducted and presented. Generative AI can automatically generate country-specific reports or forecast analyses that economists can then review, saving them valuable time. This process would have particular utility for producing annual reports or forecasting documents for policymakers, where consistency across multiple reports is essential, yet achieving this through traditional means can be tedious and error-prone. AI can thus ensure that such reports maintain a high level of precision and consistency while allowing economists to focus on the higher-order task of interpreting the results and making policy recommendations. The potential of AI to improve economic predictions is supported by empirical evidence. A study conducted by the Federal Reserve Bank of St. Louis (Jean, Bartlett, & Turcotte, 2024) demonstrated that AI models, when applied to inflation forecasting, produced lower forecast errors than traditional models across a range of countries. The ability of AI to incorporate diverse data sources, including high-frequency economic indicators and even non-traditional data such as satellite images or social media posts, enhances the robustness of these forecasts. This is particularly important in an era of rapid globalization and technological change, where traditional economic models may struggle to keep up with the pace and complexity of economic developments. Despite these advantages, the integration of AI into economic models does come with certain challenges. One of the most significant concerns is the so-called "black box" problem, where the internal workings of AI models are not easily interpretable by human experts. In economics, where transparency and theoretical justification are key, this can be a serious limitation. Economists need to be able to explain why certain predictions are being made, especially when these predictions are used to inform policy decisions that can have wide-reaching effects. To address this issue, there is growing interest in developing explainable AI models that provide not only accurate predictions but also insights into the causal relationships and mechanisms driving these predictions. This would allow economists to use AI as a tool that not only improves predictive power but also contributes to a deeper understanding of economic phenomena. Another potential issue is that AI models are highly dependent on the quality and quantity of the data they are trained on. In cases where data is limited or biased, AI models can produce misleading or skewed predictions. This is particularly relevant in economic forecasting, where the availability of reliable data can vary significantly across countries or sectors. As a result, economists must exercise caution when using AI models in contexts where data quality is a concern, ensuring that the models are calibrated and validated appropriately to avoid erroneous conclusions. Furthermore, AI’s ability to "cheat" by using future data to generate historical predictions, as identified by researchers at Desjardins (2024), illustrates the need for rigorous oversight and refinement in how AI models are deployed in economic contexts. The integration of AI into existing economic models holds great promise for improving the accuracy, efficiency, and scalability of economic predictions. By enhancing traditional models with the processing power and flexibility of AI, economists can produce more nuanced forecasts that are better suited to the complexities of a globalized world. However, to fully realize the potential of AI in economics, researchers and policymakers must continue to address challenges related to model transparency, data quality, and ethical considerations. As AI technology continues to evolve, its role in economic forecasting and analysis is likely to grow, making it an indispensable tool for both academics and practitioners. Challenging Economic Theories with Inductive Approaches The second avenue for the use of artificial intelligence in economics involves challenging traditional economic theories through inductive approaches. Inductive reasoning, which involves drawing general conclusions from specific data, is not new to economics. However, AI has the potential to elevate this method by processing vast amounts of data and identifying patterns that were previously undetectable through traditional statistical methods. This section explores how AI can be used to challenge and reshape economic theories, with a particular emphasis on its capacity to generate new hypotheses and test existing models. Inductive approaches in economics have historically been used to develop new theoretical frameworks based on empirical observations. A well-known example is the gravity model of trade, which emerged from the observation that trade between two countries is positively correlated with their economic size and negatively correlated with the distance between them. This model was not initially derived from a specific theory but from empirical regularities in the data. Similarly, AI can be employed to identify patterns and correlations in economic data that may challenge or complement existing theories. For instance, machine learning algorithms can process large datasets encompassing various economic indicators—ranging from consumer behavior to international trade flows—and reveal relationships that may not fit within the confines of traditional economic models. One of the key advantages of using AI for inductive reasoning in economics is its capacity for handling vast and complex datasets. Economic data is increasingly diverse, encompassing not only quantitative indicators like GDP or inflation rates but also qualitative data such as sentiment analysis from social media or policy documents. Traditional econometric models are often limited in their ability to incorporate these diverse data sources, as they rely on linear relationships and predefined assumptions. AI, by contrast, is capable of processing non-linear relationships and unstructured data, thus offering a more flexible framework for exploring new theoretical possibilities. This flexibility allows AI to identify new patterns in the data that may not conform to established economic theories, thereby prompting economists to rethink their assumptions and develop new hypotheses. Moreover, AI’s capacity for inductive reasoning enables economists to test existing theories in new ways. Traditionally, economic theories are tested deductively, with models being built based on theoretical assumptions and then tested against empirical data. However, AI allows for a more data-driven approach, where models are generated inductively based on patterns observed in the data, and then tested to see how well they fit with existing theories. This approach is particularly useful for exploring areas of economics where traditional models may fall short, such as the impact of technological innovation on labor markets or the dynamics of global supply chains in the context of geopolitical instability. A specific area where AI-driven inductive reasoning could have a significant impact is in macroeconomic modeling. Macroeconomic models, such as those used by central banks to forecast inflation or unemployment, are often based on a set of assumptions about how different variables interact. However, these models can be limited in their ability to account for the complexity of modern economies, where factors such as global trade dynamics, technological change, and political risk play an increasingly important role. AI offers the potential to incorporate these factors into macroeconomic models in new ways, by identifying patterns in the data that may not have been captured by traditional models. This could lead to the development of new macroeconomic theories that better reflect the realities of a rapidly changing global economy. While the potential for AI to challenge and reshape economic theories is significant, it is important to recognize the limitations of this approach. One key limitation is the "black-box" nature of many AI algorithms, which makes it difficult to understand how specific conclusions are reached. In economics, where the ability to explain and justify theoretical assumptions is critical, this lack of transparency can be a significant drawback. Economists must therefore be cautious in adopting AI-driven inductive approaches, ensuring that any new hypotheses or models generated by AI are both interpretable and theoretically sound. Furthermore, AI-driven inductive reasoning is only as good as the data on which it is based. If the data used to train AI models is biased or incomplete, the conclusions generated by the model may also be flawed. This is particularly relevant in the context of global economic data, where disparities in data availability and quality between countries can lead to skewed results. For example, countries with more comprehensive economic data may disproportionately influence the conclusions drawn by AI models, leading to theoretical frameworks that do not fully account for the experiences of less data-rich economies. AI offers significant potential to challenge and reshape traditional economic theories through inductive approaches. By processing vast amounts of diverse data and identifying new patterns, AI can prompt economists to rethink existing models and develop new hypotheses. However, the use of AI in this context must be approached with caution, given the limitations of AI algorithms and the potential for bias in the data. Future research should focus on refining AI-driven inductive approaches in economics, ensuring that the insights generated are both accurate and theoretically sound. Case Study: The Mondo Project at CIRANO The Mondo International project exemplifies a comprehensive approach to integrating artificial intelligence in economics by employing the three main families of AI: explanatory, predictive, and generative, including RAG approaches. Each of these AI categories contributes uniquely to the analysis of economic data, allowing for a more multifaceted understanding of global economic trends and decision-making. Explanatory AI, which focuses on understanding and interpreting data, is used extensively in the Mondo International project to provide insights into the causal relationships between economic variables. This branch of AI is critical for uncovering the underlying factors that drive economic outcomes, allowing economists and policymakers to move beyond mere correlations and delve into the mechanisms that explain observed patterns. In practice, explanatory AI is applied at all levels of analysis—macro, meso, and micro—helping to clarify why certain economic phenomena occur. For instance, by analyzing sectoral and regional data, explanatory AI can reveal why specific industries perform differently in response to similar economic policies or external shocks. This type of AI assists in providing clearer explanations for the complex, interconnected dynamics of global supply chains, consumer behavior, and firm-level productivity. Predictive AI, on the other hand, is central to the project’s ability to forecast future economic trends. By employing machine learning models trained on extensive datasets, predictive AI enables Mondo International to generate highly accurate forecasts for key macroeconomic indicators such as GDP growth, inflation, and unemployment. These models are not limited to traditional economic variables but incorporate real-time, high-frequency data from diverse sources, such as social media sentiment, market trends, and global news. Predictive AI allows for a more dynamic and responsive approach to economic forecasting, making it especially valuable in volatile economic environments where quick adjustments are necessary. In the Mondo International project, predictive AI models are applied not only to national economies but also to individual sectors and firms, allowing for granular, actionable forecasts that can inform both policymakers and business leaders. Generative AI represents the most innovative application of artificial intelligence within the Mondo International project. It automates the generation of reports, summaries, and even hypothetical scenarios based on the model outputs, allowing for the efficient dissemination of insights across a wide range of contexts. For example, generative AI can produce country-specific economic analyses based on the same theoretical models, providing tailored insights for different regions without requiring significant manual effort. This scalability makes generative AI an invaluable tool for organizations that need to conduct large-scale comparative analyses across multiple countries or industries. Additionally, generative AI can simulate hypothetical economic scenarios—such as the impact of policy changes or external shocks—enabling economists and policymakers to explore potential outcomes and make more informed decisions. By automating the generation of these analyses, generative AI enhances the efficiency and reach of the Mondo International project, allowing it to provide real-time insights to a global audience. The combination of these three AI families—explanatory, predictive, and generative—gives the Mondo International project a unique edge in economic analysis. Explanatory AI provides the interpretive framework necessary to understand complex economic relationships, predictive AI offers highly accurate forecasts, and generative AI automates the production and dissemination of insights. Together, these AI capabilities allow the project to tackle the challenges of modern economic analysis in a comprehensive and scalable manner. Whether analyzing macroeconomic trends, sectoral dynamics, or firm-level behavior, the Mondo International project demonstrates how AI can transform the way economists approach data, forecast outcomes, and communicate insights. The Mondo International project’s use of the three families of AI—explanatory, predictive, and generative—illustrates the profound potential of AI to reshape economic analysis. By combining these different AI techniques, the project is able to provide deeper, more accurate insights into global economic trends, while also offering scalable solutions for real-time policy analysis and decision-making. As AI technologies continue to evolve, their application in economics, as demonstrated by Mondo International, will play an increasingly central role in shaping both economic theory and practice. The primary goal of the Mondo International project is to leverage AI to address some of the long-standing challenges in economic data analysis. Economic data is notoriously complex, with different levels of aggregation and often missing, inconsistent, or outdated data points. Traditional methods for handling such data are often time-consuming and prone to error. Mondo International seeks to address these issues by employing advanced AI techniques to clean, integrate, and analyze data across multiple levels. In doing so, the project provides a more coherent and comprehensive view of global economic trends, offering insights that might otherwise be overlooked. One of the key features of the Mondo International project is its ability to analyze data at multiple levels of economic activity. At the macro level, AI is used to analyze national economic indicators such as GDP, inflation, and unemployment rates. These indicators are typically analyzed over time and across countries, allowing for the identification of global economic trends and the development of more accurate forecasting models. For example, the use of machine learning algorithms in this context allows the project to generate predictions for GDP growth across different countries, incorporating not only traditional economic indicators but also non-traditional data sources such as social media sentiment or news reports. This multidimensional analysis enables policymakers to anticipate economic shifts with greater precision. At the meso level, Mondo International focuses on sectoral data, examining how different industries or regions respond to economic policies or external shocks. This level of analysis is particularly useful for understanding the uneven effects of economic change across different parts of the economy. For instance, AI-driven analysis can reveal how a policy aimed at stimulating growth in one sector may inadvertently cause a slowdown in another. By identifying such interactions, Mondo International provides valuable insights for policymakers, allowing for more targeted and effective economic interventions. Finally, at the micro level, the project uses AI to analyze firm-level data, exploring how individual companies respond to economic conditions and policy changes. This level of analysis is particularly important for understanding the drivers of innovation and productivity within firms. AI can identify patterns in the behavior of firms that may not be immediately obvious, such as the relationship between investment in technology and long-term profitability. Additionally, by analyzing vast amounts of firm-level data, AI can help policymakers and business leaders make more informed decisions about where to allocate resources or how to support innovation. The methodologies employed by Mondo International are diverse, reflecting the complexity of the economic data being analyzed. Machine learning algorithms are used to identify patterns in the data and generate predictive models. These algorithms are trained on vast datasets, allowing them to learn from past trends and make more accurate forecasts about future economic conditions. Natural language processing (NLP) techniques are also used to analyze unstructured data, such as news articles or social media posts, which provide valuable context for understanding economic trends. For example, NLP can be used to gauge public sentiment about a particular economic policy, providing real-time feedback that can be integrated into economic models. One of the most significant findings of the Mondo International project is the identification of hidden correlations between different economic variables. By analyzing data from multiple sources and at different levels, AI has uncovered relationships that were previously unrecognized by traditional economic models. For instance, the project has identified significant links between consumer behavior in one region and industrial output in another, suggesting that global supply chains are more interconnected than previously thought. These findings have important implications for economic theory, particularly in areas such as trade and globalization, where traditional models often assume relatively simple relationships between economic variables. The implications of the Mondo International project for economics are profound. By employing AI to analyze economic data in new ways, the project is pushing the boundaries of what is possible in economic forecasting and policy analysis. The ability to integrate data from multiple sources and analyze it at different levels provides a more nuanced and accurate picture of economic conditions, allowing for more informed decision-making. This has particular relevance in the context of global economic challenges, such as the COVID-19 pandemic, where traditional economic models have struggled to account for the complexity and unpredictability of the situation. AI, with its ability to process vast amounts of data and identify hidden patterns, offers a powerful tool for navigating such challenges. The Mondo International project represents a significant advance in the application of AI to economics. By integrating AI techniques with traditional economic analysis, the project provides deeper insights into macro, meso, and micro-level economic trends. The ability to analyze data at multiple levels and from multiple sources offers a more comprehensive understanding of the global economy, with important implications for both economic theory and policy. As AI continues to evolve, projects like Mondo International will play an increasingly important role in shaping the future of economic analysis and decision-making. One of the most recent advancements in generative AI that has significant implications for economic analysis is Retrieval-Augmented Generation (RAG). RAG combines the strengths of traditional information retrieval systems, such as databases, with the generative capabilities of large language models (LLMs). This framework allows AI to access and incorporate real-time data, enhancing the contextual relevance, factual accuracy, and up-to-date nature of the generated outputs. RAG operates through a two-step process. First, it retrieves relevant information from an external knowledge base, such as web pages, databases, or proprietary economic data sources. This retrieval is facilitated by powerful search algorithms that query vast datasets to find the most relevant information. Second, this retrieved information is integrated into the LLM’s generative process, enriching the AI’s responses by providing it with a more comprehensive understanding of the topic at hand. By doing so, RAG ensures that the generated economic analyses or forecasts are not only theoretically sound but also grounded in the most current and relevant data available. This capability is particularly beneficial in the context of economic forecasting, where timely and accurate data is crucial. Traditional models often rely on historical data, which may not always reflect the rapid changes in today’s global economy. RAG, by contrast, enables models to incorporate real-time, high-frequency economic indicators such as financial market data, consumer sentiment from social media, or breaking news reports, ensuring that forecasts remain responsive to the latest developments. For example, in the Mondo International project, RAG could be used to enhance the generation of country-specific economic reports. By retrieving real-time data on inflation trends, commodity prices, or trade flows from external sources, RAG ensures that the generated reports are not only consistent with historical models but also reflect the most up-to-date global conditions. This ability to integrate and process real-time information addresses one of the main challenges of traditional AI models—relying solely on pre-trained, static knowledge bases—and offers more dynamic, data-driven insights for decision-making. RAG also helps address the "hallucination" problem often associated with generative models, where AI might generate inaccurate or misleading information. By grounding the generated outputs in factual, retrieved data, RAG promotes factual consistency, making it a particularly valuable tool in applications where accuracy is paramount, such as economic forecasting, policymaking, and financial analysis. The integration of RAG into generative AI models represents a significant step forward in enhancing the precision and reliability of AI-driven economic insights. As RAG enables AI models to access and integrate real-time data from a variety of sources, it will likely become an indispensable tool for institutions like central banks, global organizations, and policymakers who require accurate, up-to-date economic analyses to navigate the complexities of the modern global economy. Conclusion In conclusion, the integration of artificial intelligence into economics represents a significant evolution in how economic models are developed, tested, and applied. AI, through its three families—explanatory, predictive, and generative—offers a range of tools that complement and extend traditional econometric models. Explanatory AI enhances the interpretability of complex relationships within data, predictive AI improves forecasting accuracy by processing vast and diverse data sources, and generative AI automates and scales the production of economic analyses, enabling more efficient cross-country comparisons and scenario generation. These capabilities have already begun to reshape the landscape of economic research and policy-making by providing new ways to understand and anticipate global economic trends. The Mondo International project stands as an interesting case study in this broader integration of AI within economics. By applying AI techniques across multiple levels of economic data—macro, meso, and micro—Mondo International demonstrates how AI can deliver deeper, more comprehensive insights that go beyond what traditional models can offer. Its use of all three AI families allows the project to analyze large-scale data while also producing granular, specific insights for policymakers and business leaders alike. The project’s ability to scale complex economic analyses across multiple countries highlights how AI can make global economic research more efficient and accessible, thus improving decision-making on an international scale. Beyond its immediate applications, the Mondo International project also raises important questions about the future of economic theory and practice. As AI continues to evolve, projects like Mondo will challenge established economic theories by uncovering new relationships within data and generating hypotheses that could reshape the field. This inductive approach to theory development has the potential to expand our understanding of economics in ways previously unimagined, forcing researchers and policymakers to rethink their assumptions and strategies. Overall, the intersection of AI and economics is not just a technical enhancement but a paradigm shift in how we approach economic data and decision-making. Projects like Mondo International offer a glimpse into a future where AI not only improves the precision of economic forecasts but also generates entirely new ways of thinking about the economy, offering a more dynamic and data-driven approach to understanding and managing global challenges. References Jean, J., Bartlett, R., & Turcotte, S. (2024). "Inflation Forecasting Using AI: A Comparative Analysis." Federal Reserve Bank of St. Louis. Desjardins Economic Research (2024). "Point de vue économique." Desjardins, Études économiques. Turcotte, S. (2024). "Economic Forecasting and AI: A Comparative Study." HEC Montréal.