|
on Artificial Intelligence |
| By: | Pawel Struski; Jakub Swistak; Inez Okulska; Przemyslaw Biecek |
| Abstract: | Large language models (LLMs) are increasingly deployed as economic agents, yet there is little evidence whether LLM agents are suited for participating in market mechanisms designed for humans, and whether these mechanisms deliver desired outcomes when faced with LLM agents. We address this question by replicating seminal economic experiments, replacing human subjects with LLM agents. We place agents in a double auction environment, which is a widely-used market mechanism. We check whether such a market is able to deliver an efficient allocation of resources, thereby testing a novel dimension of alignment of LLM agents -- their compatibility with a fundamental market mechanism. We find that markets populated by LLM agents exhibit slower or no convergence towards market equilibrium, thus providing less efficient allocations than markets populated by humans. We then analyze agents' individual trading decisions and find substantial heterogeneity both across model families and market roles. We also run a lexical analysis of Chain-of-Thought (CoT) traces generated by the agents. We find that the decision to execute a trade rather than continue incrementally adjusting prices is associated with a shift from strategic considerations toward urgency. We publicly release our testing framework, which can be used for future evaluations. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.02580 |
| By: | Davood Wadi; Yu Ma |
| Abstract: | Search rankings are valuable because human attention is scarce and sequential. Higher-placed alternatives are easier to find, so they are examined and bought more often. Consumers are now delegating search to AI agents that can ingest an entire results page at once. Randomizing the order of one hundred hotel listings across 5, 000 AI agent sessions, we compare four large language models against human field data. AI agents search more deeply than humans and never decline to buy. Position still predicts which listings are inspected, but weakly and non-monotonically: the middle of a results page has the lowest probability of inspection, not the bottom. Position reaches the choice stage for some models and not others, a heterogeneity that tracks neither provider nor capability. All models nonetheless converge on the same undominated listing. For agentic search, the attributes displayed on a results page matter more than placement within it. |
| Date: | 2026–08 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2608.22697 |
| By: | Rumpf, Matthias; Chaliasos, Michaēl; Kosyakova, Tetyana; Otter, Thomas |
| Abstract: | This study compares belief-driven financial advice from professionals, peers, and LLM with vignettes, eliminating matching problems and enabling belief elicitation without incentive confounds. Repeated identical LLM prompts yield varied risky portfolio recommendations from shifting implicit rules. A Bayesian hierarchical Tobit model captures observed and unobserved heterogeneity. Professionals and peers respond to vignettes consistently with theory but reflect their risk preferences and characteristics. Professional advice differs in responding to client characteristics. The LLM shows smaller variance and great sensitivity to declared risk tolerance. Peers discourage stock participation among younger, lower-income investors with limited professional-advice access; AI can mitigate or reverse this. |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:imfswp:343086 |
| By: | Georgy Egorov; Konstantin Sonin |
| Abstract: | Artificial intelligence is increasingly used for political advice. We study an AI that is better informed about a payoff-relevant state and cares both about accuracy and about the perceived welfare of the individual it advises. The AI then has an incentive to tilt advice toward what the individual would like to believe, altering both the political content and the informativeness of its messages. Sophisticated individuals anticipate this distortion and filter out its predictable political component, yet still learn less because the AI makes its messages less responsive to the state. Individuals who underestimate the incentive instead mistake political accommodation for information, allowing political preferences to distort factual beliefs and generate polarization and radicalization. The model also shows that better-informed individuals receive more informative and less politically tilted advice, while greater sophistication can improve interpretation yet worsen communication itself. Contrary to the familiar echo-chamber intuition, political distortion is mitigated when political preferences and prior beliefs coincide and is most consequential when they diverge. We show how independently varying individuals’ stated preferences and prior beliefs can recover the AI’s responsiveness to the state even when the underlying state cannot be manipulated. |
| JEL: | D72 D83 P00 |
| Date: | 2026–08 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35689 |
| By: | David Autor; Tanya Rodchenko; Josh Martin; Zanna Iscenko; Scott Strand; David Pearl; Melissa Ferere |
| Abstract: | Whether AI assistance builds or erodes professional expertise is unsettled. In a pre-registered three-month randomized controlled trial, we gave 133 practicing patent lawyers at eleven U.S. intellectual property law firms access to a custom AI drafting assistant and measured both their performance while using AI and their professional judgment afterward without it. All work was scored by blinded expert patent attorneys. Paralleling findings from other white-collar domains, AI access raised the quality of work delivered on benchmark patent drafting tasks at 10 days (0.34 SD, p = 0.03) and 90 days (0.38 SD, p = 0.01), with larger gains among junior lawyers. After three months, all subjects redlined an existing patent application without AI, a core task of patent practice requiring expert judgment. Treated lawyers outperformed controls by 0.32 SD (p = 0.04), but this advantage was concentrated entirely among senior lawyers (0.45 SD, p = 0.02). Junior lawyers showed no average gain; their scores instead bifurcated, with sharply fewer mediocre scores offset by more poor and more good ones. The largest gains from AI thus accrued to the lawyers who retained the least. Foundational expertise may be a prerequisite for extracting durable skill from AI-assisted practice. |
| JEL: | I24 I29 J0 |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35720 |
| By: | Raphaël Teixeira (CERAG - Centre d'études et de recherches appliquées à la gestion - UGA - Université Grenoble Alpes, UGA - Université Grenoble Alpes, Grenoble INP - Institut polytechnique de Grenoble - Grenoble Institute of Technology - UGA - Université Grenoble Alpes) |
| Abstract: | Generative Artificial Intelligence (Gen AI) is reshaping professional practices, opening new paradigms of co-creation (Davis, 2013). Human-AI co-creation can be defined as a setting where humans and AI collaborate to create new forms of art, design, and innovation (Vinchon et al., 2023). This research investigates how Human–AI Collaboration impacts creative work. As Artificial Intelligence integration expands, it reshapes productivity, team structures, and creative dynamics across industries (Makarius, 2020). Human–AI Collaboration (HAIC) focuses on cooperative dynamics, shared goals, and team development (Fragiadakis et al., 2024) to understand how agency is built through hybrid intelligence (Sundar, 2020). Managers still lack appropriate tools to evaluate these emerging hybrid collaborations, which vary across sectors (Sauer & Burggräf, 2024). This research targets creative industries and aims to explore how co-creativity emerges as a new collaborative framework. Within the context of agentic AI, this doctoral project intends to design an interaction protocol based on Large Language Models (LLMs) to engage the team's creativity. |
| Keywords: | Multi Agent System, GenAI, Creativity, Human-AI Collaboration |
| Date: | 2026–02–02 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05723566 |
| By: | André Luduvice; Roberto Pinheiro |
| Abstract: | We model artificial intelligence (AI) as a distinct capital input in a nested CES aggregate production function that extends work done by Krusell et al. (2000) and embed it in a dynamic general equilibrium economy. By bringing the model to the data, we estimate the degree to which AI either substitutes for or complements the other production inputs at the aggregate level. We find that AI is complementary to the high-skill–equipment composite and that the AI weight in production remains small. We then use the model to study how this complementarity shapes the macroeconomic and distributional effects of AI capital accumulation. The estimated model implies different effects of alternative AI shocks. A rise in the AI usage share is contractionary as it increases reliance on a scarce complementary input. A fall in AI prices is expansionary due to the lower cost of accumulating AI capital. A tax on AI capital income raises limited revenue while the AI capital stock remains small, but can finance welfare-improving transfers as the AI price falls. A large-scale universal basic income (UBI) funded jointly by a consumption tax slows AI investment with welfare gains for low-skill workers at the expense of losses for high-skill workers and entrepreneurs. On the measurement side, we construct a quality-adjusted AI price index from hedonic regressions and build a corresponding AI capital stock. |
| JEL: | E22 E25 J24 J31 |
| Date: | 2026–09–04 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedcwq:103765 |
| By: | Zonghao Yang |
| Abstract: | Counterfactual analysis aims to predict potential outcomes under hypothetical scenarios, offering valuable insights for decision-making. This paper investigates the application of large language models (LLMs), specifically the GPT-3.5 model, for counterfactual analysis. We focus on the online lending context, where the counterfactual return on investment (ROI) is crucial for evaluating different interest rate schemes. We begin by assessing the predictive performance of GPT and comparing it with advanced machine learning algorithms. The results show that prompt engineering can significantly enhance GPT's predictions, with the R-squared increasing from 1.97% to 2.84%, closely approaching the 3.48% achieved by gradient-boosted regression. Subsequently, we utilize GPT to generate counterfactual ROIs under a set of alternative interest rates. GPT exhibits logical coherence and causal reasoning in its responses. The findings underscore the potential of LLMs as effective tools for counterfactual analysis in online lending, suggesting broader applications for LLMs in various predictive and decision-making contexts. |
| Date: | 2026–08 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2608.05367 |
| By: | Tania Babina; Alex X. He; Renhao Jiang |
| Abstract: | Using a new firm-level measure of AI investment based on AI-skilled employment—spanning machine learning through generative and agentic AI—we show that AI investments are associated with productivity growth in recent years, but not over the previous decade. We trace the productivity gains to the accumulation of organization capital that AI helps create: durable firm-specific knowledge acquired through learning-by-doing that enables more efficient production. We build a novel measure of organization capital based on workers’ job descriptions and document that productivity gains are driven by AI-skilled jobs that build organization capital. Overall, our findings suggest that AI investment generates productivity growth by creating organization capital. |
| JEL: | D22 D24 D25 E22 G3 G30 G32 J24 O33 |
| Date: | 2026–08 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35684 |
| By: | Alexander Bick; Adam Blandin; David Deming; Tyler Schumacher |
| Abstract: | We measure how workers use genAI for their jobs in a nationally representative survey linking genAI adoption to detailed occupations and tasks. Our data provide the first task-level genAI adoption indexes, which we show can inform analyses of genAI's labor market impact. Exposure scores explain some, but far from all, of the variation in adoption across occupations and tasks. We also distinguish our indexes from measures based on genAI platform chat logs, which differ conceptually and tend to over-classify chats into generic activities spanning many occupations. Finally, we highlight that current adoption is widespread but shallow: genAI is used across many occupations and tasks, yet within most of them, fewer than half of workers adopt. This indicates substantial variation among workers doing very similar work, suggesting that understanding who adopts may matter as much as understanding which tasks genAI assists. |
| Keywords: | generative artificial intelligence (AI); technology adoption; occupation and task-level measurement; artificial intelligence (AI) exposure; labor markets |
| JEL: | J24 O33 C83 |
| Date: | 2026–08–25 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedlwp:103693 |
| By: | Yi Chen; Zhe Wang; Jing Zhou |
| Abstract: | We study how AI regulation affects firm valuation using the EU Artificial Intelligence Act, the world's first comprehensive AI framework. In an event study around the April 2021 proposal, we find firms combining deeper EU presence with faster AI hiring earned higher announcement returns, suggesting markets value “jurisdictional capital”—experience in the EU regulatory environment helps firms navigate the AI regulation. The effect is stronger for high-risk AI, for firms with stable and concentrated EU presence, or prior compliance experience, unexplained by size, foreign exposure, or lobbying. EU-embedded, AI-expanding firms increase within-firm EU revenue share when peers are less embedded. |
| Keywords: | AI regulation; event study; EU AI Act; jurisdictional capital |
| Date: | 2026–08–28 |
| URL: | https://d.repec.org/n?u=RePEc:imf:imfwpa:2026/180 |
| By: | Patrick A. Imam; Jonathan R. W. Temple |
| Abstract: | Will artificial intelligence (AI) help poorer countries catch up? This paper argues that the answer depends less on access to AI than on the capacity to use new knowledge productively. We show that countries have narrowed gaps in capital and schooling more readily than gaps in productivity. Technology can diffuse widely without producing productivity convergence. Measured human capital explains only a modest share of productivity differences in levels, but is associated with sharply different mobility regimes. The estimated transition processes imply an expected time to exit the lowest-productivity state of about 65 years for economies below the estimated human-capital threshold, compared with about 25 years for those above it. This reconciles development accounting with the view of human capital as absorptive capacity. If AI mainly augments skilled workers and capable firms, it may reinforce existing gaps. If it lowers the costs of learning, adaptation, and implementation in weaker-capability economies, it could instead promote convergence. The Intelligence Divide is therefore not simply about access to AI, but about the capacity to turn knowledge into productivity. |
| Keywords: | relative development; productivity mobility; human capital; technology diffusion; artificial intelligence; convergence; absorptive capacity |
| Date: | 2026–09–04 |
| URL: | https://d.repec.org/n?u=RePEc:imf:imfwpa:2026/190 |
| By: | Jermyn Zhen Yong Bek; Zhuang Qiang Bok; Zhongtian Sun |
| Abstract: | Investment management is a high-stakes domain in which agentic AI systems must do more than generate plausible text. They must retrieve point-in-time data, assemble correct computational inputs, invoke specialized methods, and produce auditable structured outputs. We introduce FinSkillBench, an evaluation suite designed to measure whether language model agents can effectively use financial domain skills to solve investment management tasks. The benchmark spans three domains, portfolio construction, risk management, and fundamental analysis, and includes 12 subtasks with 2, 603 task episodes. Each episode provides point-in-time inputs, hidden ground truth, and a task-specific verifier.We compare three conditions: no skill, curated skill packages consisting of procedural documents and executable components, and self-generated skills in which the agent writes and reuses its own procedures within an episode. Across 9 models and a large-scale evaluation, curated skills consistently improve performance, raising mean scores from 0.366 to 0.528, with the largest gains in portfolio construction and risk management. In contrast, self-generated skills provide little benefit despite higher computational cost. An independent evaluation using a separate agent framework (Hermes Agent, 8 models, 5, 280 episodes total) reproduces the directional pattern across all three domains, with the magnitude of skill effects varying by subtask and harness. These results showthat in investment management agents, access to reliable procedural skills can be as important as model choice, while naive self-generation of skills is often ineffective. We release the benchmark, evaluation tools, curated skill packages, and full trajectories to support further research. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2608.18099 |