|
on Artificial Intelligence |
| By: | Anton Korinek (University of Virginia and Centre for the Governance of AI); Jai Vipra (University of Virginia and Centre for the Governance of AI) |
| Abstract: | This paper examines the evolving structure and competition dynamics of the rapidly growing market for foundation models, focusing on large language models (LLMs). We describe the technological characteristics that shape the industry and have given rise to fierce competition among the leading players. The paper analyzes the cost structure of foundation models, emphasizing the importance of key inputs such as computational resources, data, and talent, and identifies significant economies of scale and scope that may create a tendency towards greater market concentration in the future. We explore two concerns for competition, the risk of market tipping and the implications of vertical integration, and use our analysis to inform policy remedies to maintain a competitive landscape. |
| Keywords: | Artificial Intelligence, economic concentration, vertical integration, AI regulation. |
| JEL: | D43 O33 L86 L40 L41 K21 |
| Date: | 2024–10–02 |
| URL: | https://d.repec.org/n?u=RePEc:thk:wpaper:inetwp228 |
| By: | Cheng, Yanan (Wuhan University); Gao, Baojun (Wuhan University); Zhang, Ran (Alan) (Texas Tech University - Area of Information Systems and Quantitative Sciences (ISQS)); Li, Xitong (HEC Paris) |
| Abstract: | Many online platforms have recently integrated generative AI (GAI) generated content, yet its impact on the platform ecosystem requires careful investigation. By leveraging a unique policy of a leading online review platform that introduces GAI reviews summary (GAIRS), this study examines how GAIRS can affect the matching quality of online consumers purchasing products or services. Constructing a unique panel dataset of online reviews for a set of hotels on both TripAdvisor and Expedia, we apply a cross-platform difference-indifferences approach to estimate the impact of GAIRS. Our findings elucidate the positive effects of GAIRS on matching quality, manifested by increased consumer rating and decreased rating dispersion. This effect is driven by a decrease in unsatisfactory consumer experiences. Exploring potential mechanisms, we show that the positive effect of GAIRS on matching quality is more prominent in hotels with high uncertainty and when GAIRS is generated from a larger number of reviews, contains more content, or exhibits greater readability. We also present direct evidence supporting our mechanism by showing that the consumer reviews post-GAIRS display greater certainty and assertiveness in their content. Our further analyses rule out an alternative explanation for GAIRS's role being a form of top review, by showing evidence for the performance of solicited reviews, the absence of consumer imitation from GAIRS, and improvements in hotel performance. Finally, we employ transfer deep learning to further demonstrate that GAIRS can reduce uncertainty. Additionally, we find that improvements in experiential dimensions including rooms, value, noise level, and service drive the decline in unsatisfactory consumer experiences. This research highlights the potential of GAIRS, as a recent GAIempowered application in online platforms, in improving matching between online consumers and products, thereby contributing to the expanding discourse on the impacts of GAI in online markets. |
| Keywords: | Generative AI reviews summary; matching quality; natural experiment; difference-in-differences; uncertainty reduction |
| JEL: | C88 |
| Date: | 2025–02–04 |
| URL: | https://d.repec.org/n?u=RePEc:ebg:heccah:1545 |
| By: | Abbas Nejad, Kian (Tilburg University, Center For Economic Research); Musillo, Giuseppe (Tilburg University, Center For Economic Research); Wicker, Till (Tilburg University, Center For Economic Research); Zaccaria, Niccolò (Tilburg University, Center For Economic Research) |
| Keywords: | large language models; cover letters; labor market; matching; signaling |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:tiu:tiucen:a808b376-b6ae-4b5f-8a35-57fe320619eb |
| By: | Raphael Auer; David Kopfer; Josef Sveda |
| Abstract: | How exposed is the labour market to ever-advancing AI capabilities, to what extent does this substitute human labour, and how will it affect inequality? We address these questions in a simulation of 711 US occupations classified by the importance and level of cognitive skills. We base our simulations on the notion that AI can only perform skills that are within its capabilities and involve computer interaction. At low AI capabilities, 7% of skills are exposed to AI uniformly across the wage spectrum. At moderate and high AI capabilities, 17% and 36% of skills are exposed on average, and up to 45% in the highest wage quartile. Examining complementary versus substitution, we model the impact on side versus core occupational skills. For example, AI capable of bookkeeping helps doctors with administrative work, freeing up time for medical examinations, but risks the jobs of bookkeepers. We find that low AI capabilities complement all workers, as side skills are simpler than core skills. However, as AI capabilities advance, core skills in lower-wage jobs become exposed, threatening substitution and increased inequality. In contrast to the intuitive notion that the rise of AI may harm white-collar workers, we find that those remain safe longer as their core skills are hard to automate. |
| Keywords: | Artificial intelligence, automation, chatGPT, employment, GPT, inequality, labour market, LLM, technology, wage |
| JEL: | E24 E51 G21 G28 J23 J24 M48 O30 O33 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:cnb:wpaper:2025/6 |
| By: | Karol Madoń |
| Abstract: | This paper studies the relationship between exposure to artificial intelligence (AI) and workers’ wages across European countries. Overall, a positive relationship between exposure to AI and workers’ wages is found, however it differs considerably between workers and countries. High-skilled workers experience far higher wage premiums related to AI-related skills than middle- and low-skilled workers. Positive associations are concentrated among occupations moderately and highly exposed to AI (between the 6th and 9th decile of the exposure), and are weaker among the least exposed occupations. Returns of AI-related skills among high-skilled workers are even higher in Eastern European Countries compared to Western European countries. The heterogeneity likely originates from the difference in overall labour costs between country groups. The results presented in this study were obtained from the estimation of Mincerian wage regressions on the 2018 release of the EU Structure of Earning Survey. |
| Keywords: | artificial intelligence, wages, technological change, Europe |
| JEL: | E24 J30 O33 |
| Date: | 2025–03 |
| URL: | https://d.repec.org/n?u=RePEc:ibt:wpaper:wp052024 |
| By: | Sergio Galeano; Nye Hodge; Alexander Ruder |
| Abstract: | The rapid advancement of artificial intelligence (AI) has prompted widespread interest and discussion about its potential to transform the labor market. For workforce development practitioners, a key issue is how AI is changing the nature of work, mainly through changes in the skills workers need to be competitive for the jobs of today and of the future. In this Workforce Currents, we explore the growth of employer demand for AI skills in online job postings data between 2010 and 2024. Lightcast, a labor analytics firm, provides job postings data that includes several useful features of each posting, such as the skills required to perform the job functions, the education level required, and the occupation title. While other researchers have documented the overall and industry- or occupation-specific growth in AI skill demand, we investigate how AI skill demand has changed for occupations with differing levels of required education. We ask whether the growth in AI skill demand is concentrated in occupations that require a bachelor’s degree or higher, or whether AI skill demand is growing even in occupations that require an associate degree or high school diploma. The answer to these questions can inform workers as they choose training programs and help workforce development practitioners align certificate and associate degree curricula to meet the changing needs of employers. |
| Keywords: | Artificial intelligence; Workforce trends; workforce changes; workforce training |
| JEL: | J21 J23 J24 O33 |
| Date: | 2025–05–21 |
| URL: | https://d.repec.org/n?u=RePEc:fip:a00034:100013 |
| By: | Yusuke Aoki; Joon Suk Park; Yuya Takada; Koji Takahashi |
| Abstract: | This paper examines the relationship between individuals' expectations of job replacement by generative AI (GenAI) and their macroeconomic outlooks and behaviors. Using online surveys combined with randomized experiments conducted in the U.S. and Japan, we derive the following findings about the effects of expecting greater job replacement due to GenAI. First, in both the U.S. and Japan, respondents revise their beliefs after receiving information about GenAI's job replacement ratios. Second, in Japan, such an expectation leads to an increase in inflation expectations driven by a rise in investment. Third, it increases respondents' willingness to use GenAI in workplaces in Japan. Fourth, in the U.S., expectations of greater job replacement amplify concerns about weaker short-term labor demand and reduced skill requirements, particularly among more educated respondents. In addition, these respondents anticipate lower investment, while less educated respondents expect higher investment. |
| Keywords: | Generative Artificial intelligence, labor market, inflation, productivity |
| JEL: | E24 E31 O30 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:bis:biswps:1269 |
| By: | König, Pascal; Arnesen, Sveinung |
| Abstract: | This article tests the argument that the reliance on AI systems affects people’s affective ties to government employees using AI systems. Drawing on social cognition theory, it examines how AI use influences the perceived warmth of public servants and the acceptability of decision-making. It distinguishes between two settings in the education system that differ regarding how directly citizens are affected by AI use, a teacher using AI to help assess students and a public servant allocating funds among schools. The analysis is based on a pre-registered vignette experiment and a sample of 4, 569 participants from Norway. It finds that AI use decreases both the perceived warmth and competence of public servants, that these evaluations negatively bear on the overall acceptability of decision-making, and that the effect of AI use is stronger for public servants more directly interacting with citizens. The findings have important implications for the legitimacy of public organizations. |
| Date: | 2025–05–03 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:732ez_v1 |
| By: | Qiang Chen; Tianyang Han; Jin Li; Ye Luo; Yuxiao Wu; Xiaowei Zhang; Tuo Zhou |
| Abstract: | Can AI effectively perform complex econometric analysis traditionally requiring human expertise? This paper evaluates an agentic AI's capability to master econometrics, focusing on empirical analysis performance. We develop an ``Econometrics AI Agent'' built on the open-source MetaGPT framework. This agent exhibits outstanding performance in: (1) planning econometric tasks strategically, (2) generating and executing code, (3) employing error-based reflection for improved robustness, and (4) allowing iterative refinement through multi-round conversations. We construct two datasets from academic coursework materials and published research papers to evaluate performance against real-world challenges. Comparative testing shows our domain-specialized agent significantly outperforms both benchmark large language models (LLMs) and general-purpose AI agents. This work establishes a testbed for exploring AI's impact on social science research and enables cost-effective integration of domain expertise, making advanced econometric methods accessible to users with minimal coding expertise. Furthermore, our agent enhances research reproducibility and offers promising pedagogical applications for econometrics teaching. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.00856 |
| By: | Kim Kaivanto |
| Abstract: | In policy debates concerning the governance and regulation of Artificial Intelligence (AI), both the Precautionary Principle (PP) and the Innovation Principle (IP) are advocated by their respective interest groups. Do these principles offer wholly incompatible and contradictory guidance? Does one necessarily negate the other? I argue here that provided attention is restricted to weak-form PP and IP, the answer to both of these questions is “No.†The essence of these weak formulations is the requirement to fully account for type-I error costs arising from erroneously preventing the innovation’s diffusion through society (i.e. mistaken regulatory redlighting) as well as the type-II error costs arising from erroneously allowing the innovation to diffuse through society (i.e. mistaken regulatory green-lighting). Within the Signal Detection Theory (SDT) model developed here, weak-PP red-light (weak-IP green-light) determinations are optimal for sufficiently small (large) ratios of expected type-I to type-II error costs. For intermediate expected cost ratios, an amber-light ‘wait-and-monitor’ policy is optimal. Regulatory sandbox instruments allow AI testing and experimentation to take place within a structured environment of limited duration and societal scale, whereby the expected cost ratio falls within the ‘wait-and-monitor’ range. Through sandboxing regulators and innovating firms learn more about the expected cost ratio, and what respective adaptations — of regulation, of technical solution, of business model, or combination thereof, if any — are needed to keep the ratio out of the weak-PP red-light zone. |
| Keywords: | artificial intelligence, foundational AI, general-purpose AI systems, AI governance, precautionary principle, innovation principle, countervailing risk, scientific uncertainty, signal detection theory, misclassification costs, discriminability, ROC curve, de minimis risk, trust and polarization, protected values, non-comparable values, continuity axiom, regulatory sandboxes |
| JEL: | D81 O31 O33 O38 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:lan:wpaper:423283411 |
| By: | Hongyang Yang; Likun Lin; Yang She; Xinyu Liao; Jiaoyang Wang; Runjia Zhang; Yuquan Mo; Christina Dan Wang |
| Abstract: | Enterprise Resource Planning (ERP) systems serve as the digital backbone of modern financial institutions, yet they continue to rely on static, rule-based workflows that limit adaptability, scalability, and intelligence. As business operations grow more complex and data-rich, conventional ERP platforms struggle to integrate structured and unstructured data in real time and to accommodate dynamic, cross-functional workflows. In this paper, we present the first AI-native, agent-based framework for ERP systems, introducing a novel architecture of Generative Business Process AI Agents (GBPAs) that bring autonomy, reasoning, and dynamic optimization to enterprise workflows. The proposed system integrates generative AI with business process modeling and multi-agent orchestration, enabling end-to-end automation of complex tasks such as budget planning, financial reporting, and wire transfer processing. Unlike traditional workflow engines, GBPAs interpret user intent, synthesize workflows in real time, and coordinate specialized sub-agents for modular task execution. We validate the framework through case studies in bank wire transfers and employee reimbursements, two representative financial workflows with distinct complexity and data modalities. Results show that GBPAs achieve up to 40% reduction in processing time, 94% drop in error rate, and improved regulatory compliance by enabling parallelism, risk control insertion, and semantic reasoning. These findings highlight the potential of GBPAs to bridge the gap between generative AI capabilities and enterprise-grade automation, laying the groundwork for the next generation of intelligent ERP systems. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.01423 |