|
on Artificial Intelligence |
| By: | Hadi Hosseini; Samarth Khanna; Ronak Singh |
| Abstract: | The rise of Large Language Models (LLMs) has driven progress in reasoning tasks -- from program synthesis to scientific hypothesis generation -- yet their ability to handle ranked preferences and structured algorithms in combinatorial domains remains underexplored. We study matching markets, a core framework behind applications like resource allocation and ride-sharing, which require reconciling individual ranked preferences to ensure stable outcomes. We evaluate several state-of-the-art models on a hierarchy of preference-based reasoning tasks -- ranging from stable-matching generation to instability detection, instability resolution, and fine-grained preference queries -- to systematically expose their logical and algorithmic limitations in handling ranked inputs. Surprisingly, even top-performing models with advanced reasoning struggle to resolve instability in large markets, often failing to identify blocking pairs or execute algorithms iteratively. We further show that parameter-efficient fine-tuning (LoRA) significantly improves performance in small markets, but fails to bring about a similar improvement on large instances, suggesting the need for more sophisticated strategies to improve LLMs' reasoning with larger-context inputs. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.04478 |
| By: | Yong Suk Lee |
| Abstract: | This paper examines the racial implications of police interaction with algorithms, particularly in the context of racial disparities in rearrest predictions. Our experimental study involved showing police officers the profiles of young offenders and asking them to predict rearrest probabilities within three years, first without and then after seeing the algorithm’s assessment. The experiment varied the visibility of the offender’s race (revealed to one group, hidden in another group, and mixed (some shown and some hidden) in the other group). Additionally, we explored how informing officers about the model’s accuracy affected their responses. Our findings indicate that officers adjust their predictions towards the algorithm’s assessment when the race of the profile is disclosed. However, these adjustments exhibit significant racial disparities, with a significant gap in initial rearrest predictions between Black and White offenders even when all observable characteristics are controlled for. Furthermore, only Black officers significantly reduced their predictions after viewing the the algorithm’s assessments, while White officers did not. Our findings reveal the limited and nuanced effectiveness of algorithms in reducing bias in recidivism predictions, underscoring the complexities of algorithm-assisted human judgment in criminal justice. |
| Keywords: | human-computer interaction, artificial intelligence, algorithmic prediction, racial bias, criminal justice |
| JEL: | C10 D63 K40 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_11877 |
| By: | Green, Jane; Grant, Zack; Evans, Geoffrey; Inglese, Gaetano |
| Abstract: | The rapid expansion of Artificial Intelligence (AI) in the workplace has significant political implications. How can we understand perceptions of both personal job risks and opportunities, given each may affect political attitudes differently? We use an original, representative survey from Great Britain to reveal; (i) the degree to which people expect personal AI-based occupational risks versus opportunities, (ii) how much this perceived exposure corresponds to variation in existing expert-derived occupational AI-exposure measures; (iii) the social groups who expect to be AI winners and AI losers; and (iv) how personal AI expectations are associated with demand for different political policies. We find that over 1-in-3 British workers anticipate being an AI winner (10%) or loser (24%) and, while expectations correlate with classifications of occupational exposure, factors like education, gender, age, and employment sector also matter. Politically, both self-anticipated AI winners and losers show similar support for redistribution, but they differ on investment in education and training as well as on immigration. Our findings emphasise the importance of considering subjective winners and losers of AI; these patterns cannot be explained by existing occupational classifications of AI exposure. |
| Keywords: | economic insecurity; welfare state; public opinion; Artificial intelligence; job prospects; technology |
| JEL: | R14 J01 J1 |
| Date: | 2025–06–30 |
| URL: | https://d.repec.org/n?u=RePEc:ehl:lserod:128289 |
| By: | Anders Humlum; Emilie Vestergaard |
| Abstract: | We examine the labor market effects of AI chatbots using two large-scale adoption surveys (late 2023 and 2024) covering 11 exposed occupations (25, 000 workers, 7, 000 workplaces), linked to matched employer-employee data in Denmark. AI chatbots are now widespread—most employers encourage their use, many deploy in-house models, and training initiatives are common. These firm-led investments boost adoption, narrow demographic gaps in take-up, enhance workplace utility, and create new job tasks. Yet, despite substantial investments, economic impacts remain minimal. Using difference-in-differences and employer policies as quasi-experimental variation, we estimate precise zeros: AI chatbots have had no significant impact on earnings or recorded hours in any occupation, with confidence intervals ruling out effects larger than 1%. Modest productivity gains (average time savings of 3%), combined with weak wage pass-through, help explain these limited labor market effects. Our findings challenge narratives of imminent labor market transformation due to Generative AI. |
| JEL: | J23 J24 J31 O33 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33777 |
| By: | Eleanor W. Dillon; Sonia Jaffe; Nicole Immorlica; Christopher T. Stanton |
| Abstract: | We present evidence on how generative AI changes the work patterns of knowledge workers using data from a 6-month-long, cross-industry, randomized field experiment. Half of the 7, 137 workers in the study received access to a generative AI tool integrated into the applications they already used for emails, document creation, and meetings. We find that access to the AI tool during the first year of its release primarily impacted behaviors that workers could change independently and not behaviors that require coordination to change: workers who used the tool in more than half of the sample weeks spent 3.6 fewer hours, or 31% less time on email each week (intent to treat estimate is 1.3 hours) and completed documents moderately faster, but did not significantly change time spent in meetings. |
| JEL: | L23 M1 M15 M5 O33 |
| Date: | 2025–05 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33795 |
| By: | Matthias Oschinski; Christian Spielmann; Sonali Subbu-Rathinam |
| Abstract: | Similar to electricity and the steam engine, Artificial intelligence (AI) is considered to be a general-purpose technology (Crafts, 2021). As such, it holds the potential to transform numerous occupations and industries, reshaping the career trajectories of many students, including those studying economics. This paper examines skills requirements for jobs most commonly pursued by U.S. economics graduates, analyses how skills demand has already changed between 2015 and 2023, a period of rapid AI development, and discusses AI’s likely impact on job profiles. Based on our findings, we explore the implications for economics teaching at the university level. Adapting university curricula to these labor market shifts and equipping students with relevant skills is crucial for shaping a future-ready economics education. |
| Date: | 2025–04–02 |
| URL: | https://d.repec.org/n?u=RePEc:bri:uobdis:25/788 |
| By: | Wynne, Richard; Kolachalama, Vijaya B |
| Abstract: | Scholarly publishing is a multibillion-dollar industry that evaluates research outputs and helps curate researcher reputation. These functions are underpinned by more than 100 million hours of volunteer peer reviewer time, representing an estimated $1.5 billion just in the United States . However, the status quo is under pressure from multiple forces: a growing number of research publications requiring peer review, questions about research integrity and reproducibility , business model changes and legal challenges from unpaid peer reviewers. The emergence of AI has exacerbated the problems, but AI also represents an opportunity to ameliorate the peer review experience and improve standards. We propose a framework for integrating AI into scholarly peer review that leverages the strengths of AI and human expertise. The framework offers a structured approach for evaluating the application of AI, with the aim of improving efficiency, consistency, comprehensiveness, and quality in the evaluation of scholarly contributions. |
| Date: | 2025–05–16 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:s764u_v1 |
| By: | Martin Elias De Simone; Federico Hernan Tiberti; Maria Rebeca Barron Rodriguez; Federico Alfredo Manolio; Wuraola Mosuro; Eliot Jolomi Dikoru |
| Abstract: | This study evaluates the impact of a program leveraging large language models for virtual tutoring in secondary education in Nigeria. Using a randomized controlled trial, the program deployed Microsoft Copilot (powered by GPT-4) to support first-year senior secondary students in English language learning over six weeks. The intervention demonstrated a significant improvement of 0.31 standard deviation on an assessment that included English topics aligned with the Nigerian curriculum, knowledge of artificial intelligence and digital skills. The effect on English, the main outcome of interest, was of 0.23 standard deviations. Cost-effectiveness analysis revealed substantial learning gains, equating to 1.5 to 2 years of ’business-as-usual’ schooling, situating the intervention among some of the most cost-effective programs to improve learning outcomes. An analysis of heterogeneous effects shows that while the program benefits students across the baseline ability distribution, the largest effects are for female students, and those with higher initial academic performance. The findings highlight that artificial intelligence-powered tutoring, when designed and used properly, can have transformative impacts in the education sector in low-resource settings. |
| Date: | 2025–05–19 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:wbrwps:11125 |
| By: | Muhammed Golec; Maha AlabdulJalil |
| Abstract: | Large Language Models (LLM), which have developed in recent years, enable credit risk assessment through the analysis of financial texts such as analyst reports and corporate disclosures. This paper presents the first systematic review and taxonomy focusing on LLMbased approaches in credit risk estimation. We determined the basic model architectures by selecting 60 relevant papers published between 2020-2025 with the PRISMA research strategy. And we examined the data used for scenarios such as credit default prediction and risk analysis. Since the main focus of the paper is interpretability, we classify concepts such as explainability mechanisms, chain of thought prompts and natural language justifications for LLM-based credit models. The taxonomy organizes the literature under four main headings: model architectures, data types, explainability mechanisms and application areas. Based on this analysis, we highlight the main future trends and research gaps for LLM-based credit scoring systems. This paper aims to be a reference paper for artificial intelligence and financial researchers. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.04290 |
| By: | Jizhou Wang; Xiaodan Fang; Lei Huang; Yongfeng Huang |
| Abstract: | Economic inequality is a global challenge, intensifying disparities in education, healthcare, and social stability. Traditional systems like the U.S. federal income tax reduce inequality but lack adaptability. Although models like the Saez Optimal Taxation adjust dynamically, they fail to address taxpayer heterogeneity and irrational behavior. This study introduces TaxAgent, a novel integration of large language models (LLMs) with agent-based modeling (ABM) to design adaptive tax policies. In our macroeconomic simulation, heterogeneous H-Agents (households) simulate real-world taxpayer behaviors while the TaxAgent (government) utilizes LLMs to iteratively optimize tax rates, balancing equity and productivity. Benchmarked against Saez Optimal Taxation, U.S. federal income taxes, and free markets, TaxAgent achieves superior equity-efficiency trade-offs. This research offers a novel taxation solution and a scalable, data-driven framework for fiscal policy evaluation. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.02838 |
| By: | Konstantin Boss; Luigi Longo; Luca Onorante |
| Abstract: | Using a state-of-the-art large language model, we extract forward-looking and context-sensitive signals related to inflation and unemployment in the euro area from millions of Reddit submissions and comments. We develop daily indicators that incorporate, in addition to posts, the social interaction among users. Our empirical results show consistent gains in out-of-sample nowcasting accuracy relative to daily newspaper sentiment and financial variables, especially in unusual times such as the (post-)COVID-19 period. We conclude that the application of AI tools to the analysis of social media, specifically Reddit, provides useful signals about inflation and unemployment in Europe at daily frequency and constitutes a useful addition to the toolkit available to economic forecasters and nowcasters. |
| Date: | 2025–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2506.10546 |