# q-fin.EC Published articles for q-fin.EC. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Human Responses to Algorithmic Correction: Evidence from Centre Court DevFeed: [Human Responses to Algorithmic Correction: Evidence from Centre Court]() Original publisher: [Read original article]() Author: David Almog, Romain Gauriot, Lionel Page, Daniel Martin Published: 2026-11-06T05:00:00Z Content type: article Language: en Sources: [arXiv: Machine Learning]() Topics: [Machine learning](), [human review]() Tags: [court](), [cs-cy](), [cs-lg](), [econ-gn](), [errors](), [framework](), [investment](), [organizational](), [q-fin-ec](), [remedy](), [responses](), [review](), [shift](), [workers]() ### AI overview Using Hawk-Eye records from professional tennis, the study finds that near the line (within 20 mm), mistaken "in" calls rise by 8.6 percentage points while mistaken "out" calls fall by 4.1 points. Within matches, calls shift toward avoiding errors that were previously overturned. The authors find no evidence that initial accuracy worsens within 100 mm of the line, despite concerns that workers might rely on review as a safety net. Their investment framework combines human responses with correction benefits and remedy costs, suggesting that estimates of review's organizational returns can be misleading if they assume human decisions stay fixed. ### Source excerpt arXiv:2401.16754v5 Announce Type: replace Abstract: We study how algorithmic review changes error composition using Hawk-Eye records from professional tennis. Within 20 mm of the line, mistaken "in" call rates increase by 8.6 percentage points, while mistaken "out" call rates decrease by 4.1 points. Within matches, calls shift toward avoiding previously overturned errors. We find no evidence that initial accuracy deteriorates within 100 mm of the line, despite concerns that workers may rely on correction as a safety net. We derive an investment framework incorporating human responses alongside correction benefits and remedy costs, showing that holding human decisions fixed can misstate the organizational returns to review. ## Does Frontier AI Strengthen Cyber Defense? Evidence from Project Glasswing DevFeed: [Does Frontier AI Strengthen Cyber Defense? Evidence from Project Glasswing]() Original publisher: [Read original article]() Author: Thomas Mosk Published: 2026-11-05T05:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [Frontier AI](), [code security](), [cve remediation]() Tags: [ai](), [claude-mythos](), [claude-mythos-preview](), [cs-ai](), [cyber-defense](), [defense](), [disclosure](), [disclosures](), [econ-gn](), [exploited](), [external](), [fixed](), [frontier-ai](), [impact](), [project-glasswing](), [q-fin-ec](), [software](), [software-security](), [software-vulnerabilities](), [vulnerabilities]() ### AI overview Using publicly disclosed vulnerabilities, the study finds that firms given early access to Claude Mythos Preview through Project Glasswing more than doubled disclosures compared with Chinese and Russian firms without access, and disclosed about 150% more high-impact flaws while fixing reported flaws just as quickly. External researchers shifted toward reporting more flaws in other firms' code, without leaving the market. The number of vulnerabilities exploited did not change; the share exploited within a month of disclosure fell by about two thirds, mainly because fewer were already exploited at disclosure. The author concludes that restricted access may create a security asymmetry between firms in different countries. ### Source excerpt arXiv:2611.06629v1 Announce Type: cross Abstract: Defenders and attackers race to find the same software flaws. I use the launch of Project Glasswing, which gave a small group of large US tech firms early access to Claude Mythos Preview, a model for finding software vulnerabilities, to study how access to frontier AI affects defenders, external researchers, and attackers. Using the universe of publicly disclosed vulnerabilities, I find that firms with access more than doubled their disclosures relative to Chinese and Russian tech firms without access, disclosed about 150 percent more high-impact vulnerabilities, and fixed reported flaws as quickly as before. As treated firms found more flaws themselves, external researchers, who sell their findings through bug bounties, reported about half as many vulnerabilities in treated firms' code and more in other firms' code, without leaving the market. There is no evidence that attackers exploited the additional vulnerabilities: the share exploited within a month of disclosure fell by about two thirds, mainly because a smaller share was already exploited when disclosed, while the number exploited did not change. Frontier AI helped defenders find and fix more flaws before attackers exploited them, but its restricted release creates an asymmetry in software security between firms in different countries. ## The Digital-Energy Recomposition of the Kitchin Cycle: Exploratory Evidence from U.S. IT and Irish Data-Centre Series (1995-2025) DevFeed: [The Digital-Energy Recomposition of the Kitchin Cycle: Exploratory Evidence from U.S. IT and Irish Data-Centre Series (1995-2025)]() Original publisher: [Read original article]() Author: Bartosz Ziollko, Adam Niedzialkowski, Bartosz Szyma Published: 2026-11-04T05:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [Sustainable Computing]() Tags: [2025](), [adjustment](), [aggregate](), [ai](), [assets](), [boundary](), [capacity](), [cloud](), [compute](), [correlation](), [cs-ai](), [data](), [data-centre](), [decoupling](), [digital](), [econ-gn](), [electricity](), [energy](), [hyperscaler](), [it](), [physics-soc-ph](), [q-fin-cp](), [q-fin-ec](), [q-fin-gn](), [recomposition](), [series]() ### AI overview The analysis reports that the U.S. IT-GDP correlation has a statistically significant break at 2014Q3, falling from +0.45 to −0.12; the break predates the imposed Cloud-AI boundary, which is not significant after adjustment. In Ireland, data-centre electricity correlates with Real GDP during the AI era (r=+0.41, N=16), but the relationship falls to about zero using Real GNP, consistent with shared multinational tax effects. Cross-correlations also indicate GDP tends to precede data-centre load, making compute capacity coincident or moderately lagging rather than a leading indicator. The authors describe the findings as exploratory given small AI-era samples and the lack of quarterly GNI* data. ### Source excerpt arXiv:2610.33122v1 Announce Type: cross Abstract: Using quarterly FRED and CSO Ireland data over 1995Q1-2025Q4, we examine whether U.S.\ IT investment and Irish data-centre electricity are recomposing the Kitchin cycle around energy- and compute-backed assets. Three findings stand out. First, a Quandt-Andrews sup-Wald test locates a significant break in the U.S.\ IT-GDP correlation at 2014Q3 (sup-W $= 9.69$; $r$ falls from $+0.45$ to $-0.12$); the Fisher $r$-to-$z$ test survives Bonferroni adjustment ($z = +3.11$, $p_{\text{Bonf}} = 0.006$), whereas the analogous test on the imposed Cloud-vs-AI boundary does not ($p_{\text{Bonf}} = 0.164$), so ``Strategic Decoupling'' pre-dates the AI Era by approximately six years. Second, in Ireland the synchronous AI-Era correlation between metered data-centre electricity and Real GDP is large ($r = +0.41$, $N = 16$) but \emph{collapses to $r \approx 0$ when Real GDP is replaced by Real GNP}, suggesting the headline is largely an artefact of the shared multinational tax-structure footprint between hyperscaler operations and Irish gross output. Third, the Irish cross-correlation profile is asymmetric in the opposite direction to a leading-indicator reading: the largest positive coefficients cluster on the $k < 0$ side (GDP precedes data-centre load), so compute capacity is better read as coincident or moderately lagging. Given the small AI-Era samples and the absence of a quarterly multinational-adjusted aggregate such as GNI*, all coefficients are exploratory; subject to these caveats, the U.S.\ Kitchin cycle is best dated to 2014Q3, and the Irish Energy-Data Nexus is dominated by multinational firm activity. ## A Financial Brain Scan of the LLM DevFeed: [A Financial Brain Scan of the LLM]() Original publisher: [Read original article]() Author: Hui Chen, Antoine Didisheim, Mo Pourmohammadi, Luciano Somoza, Hanqing Tian Published: 2026-11-04T05:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [Large Language Model](), [model architecture]() Tags: [autoencoders](), [causal-inference](), [concepts](), [cs-ai](), [cs-ce](), [econ-gn](), [embeddings](), [features](), [financial](), [language-models](), [lightweight](), [llm](), [mapping](), [q-fin-ec](), [q-fin-gn](), [quantitative-finance](), [reasoning](), [steering]() ### AI overview The abstract reports that sparse autoencoders (SAEs) can map LLM activations to interpretable concepts and steer selected features. Prompting models to express preferences also activates unintended concepts, making causal inference harder; SAE steering is described as smoother and more isolated. The scans find a disconnect between internal feature contributions and models' self-generated explanations, which the authors say are unreliable guides to reasoning. SAEs are also presented as a lightweight intervention for existing models that can produce interpretable embeddings and attenuate specific economic biases. ### Source excerpt arXiv:2508.21285v3 Announce Type: replace-cross Abstract: We use sparse autoencoders (SAEs) to "brain scan" large language models, mapping internal activations to interpretable concepts and enabling targeted "steering" of specific features. We show that prompting models to adopt specific preferences activates many unintended concepts, complicating causal inference. In contrast, SAEs permit smooth, more isolated steering of target concepts. Furthermore, our scans reveal a stark disconnect between the model's internal feature contributions and its self-generated explanations, showing that these explanations are unreliable guides to its reasoning. Finally, SAEs provide a lightweight intervention for existing models, producing interpretable embeddings and allowing researchers to attenuate specific economic biases. ## Designing Behavioral Profiles for Language Model Simulation of Out-of-Distribution Human Behavior DevFeed: [Designing Behavioral Profiles for Language Model Simulation of Out-of-Distribution Human Behavior]() Original publisher: [Read original article]() Author: Robin Na, Kehang Zhu, Logan Cross, Michiel A. Bakker Published: 2026-11-03T05:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [Large Language Model](), [Simulation](), [LLM evaluation / benchmarking](), [generalization in machine learning](), [Artificial Intelligence]() Tags: [behavior](), [cs-ai](), [distribution](), [econ-gn](), [feature-selection](), [features](), [framework](), [game](), [language-model](), [people](), [q-fin-ec](), [simulation]() ### AI overview The framework tests which behavioral descriptions help LLMs predict human behavior under new experimental designs. One LLM builds profiles from earlier participants' actions, messages, and surveys; a separate LLM simulates behavior with those profiles. Candidate features are selected by adding them when they improve predictions on learning data, then evaluated unchanged on new participants and designs. Across public-goods games, multi-party bargaining, and generalized 11-20 money-request designs, leading one-feature profiles predicted outcome distributions better than no profile or general-purpose personas. Adding features showed no clear further benefit. Features' relative predictive value largely persisted between learning and validation, but transferring profiles across game families produced uneven gains and could reduce accuracy. ### Source excerpt arXiv:2611.00661v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used to simulate human behavior, but which descriptions of people improve prediction when experimental designs change remains unclear. We develop a framework that draws candidate behavioral features from the literature and constructs profiles from earlier human experiments. A profile-generation LLM assigns feature values from observed actions, messages, and surveys; a separate LLM simulates behavior using these profiles. By comparing simulated and observed outcomes in the learning data, we select profile contents through forward feature selection, adding features while they improve prediction. We then evaluate the selected contents without modification against outcomes from new participants under new designs. Across 616 public-goods games, 144 multi-party bargaining games, and 1,480 generalized 11-20 money-request designs, leading one-feature profiles predict human outcome distributions better than no profile or general-purpose personas. Profiles with more features provide no clear additional benefit, and features' relative predictive value largely persists across learning and validation. Using profiles from one game family to simulate another produces uneven gains and can make predictions less accurate than simulations without profiles. The framework makes the choice of behavioral information a testable part of simulation design, helping researchers assess which descriptions remain useful as experimental conditions change. ## Strategic Integration of Artificial Intelligence in the C-Suite: The Role of the Chief AI Officer DevFeed: [Strategic Integration of Artificial Intelligence in the C-Suite: The Role of the Chief AI Officer]() Original publisher: [Read original article]() Author: Marc Schmitt Published: 2026-10-30T04:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [AI Strategy](), [ai-governance](), [Artificial Intelligence]() Tags: [ai](), [ai-governance](), [artificial-intelligence](), [c-suite](), [cs-ai](), [cs-cy](), [cs-lg](), [econ-gn](), [executive-leadership](), [governance](), [integration](), [organization](), [q-fin-ec](), [roles]() ### AI overview The paper explains why organizations choose different structures for integrating AI into executive leadership. It identifies three features of AI systems--distributed accountability for judgment, upstream governance, and non-stationarity--and proposes three responses: concentrated extension of an existing mandate, distributed extension across roles, or creation of a dedicated Chief AI Officer (CAIO). Four propositions address when a CAIO emerges, which configuration an organization adopts, when that role is effective, and how configurations change. An archival census finds that executive AI leadership has spread sharply since 2023; role creation and mandate extension are advancing at nearly equal rates, adoption is concentrated where digital leadership structures already exist, and executive tenures challenge claims that the CAIO is a transient role. ### Source excerpt arXiv:2407.10247v4 Announce Type: replace-cross Abstract: The integration of Artificial Intelligence (AI) into corporate strategy has become critical for organizations seeking to maintain competitive advantage in the digital age. Although organizations increasingly rely on AI as a strategic and organizational resource, existing C-suite roles remain only partially equipped to govern, integrate, and leverage it coherently at the enterprise level. Organizations vary in their responses. Some create a dedicated Chief AI Officer (CAIO), others extend existing mandates into hybrid roles, and still others coordinate AI through federated structures. This paper develops a role-design theory to explain this variation. I identify three properties that distinguish AI systems from earlier cross-cutting enterprise technologies, distributed accountability for judgment, upstream governance, and non-stationarity, and three configurations through which organizations respond: concentrated extension, distributed extension, and role creation. Four propositions specify when a dedicated CAIO emerges, what form an organization's response takes, when the role is effective, and how configurations evolve. Drawing on the first archival census of executive AI leadership, I find that the role has diffused sharply since 2023, that role creation and mandate extension are advancing in near parity, that adoption concentrates where digital leadership structures already exist, and that executive tenures contradict practitioner accounts of a transient role. The paper contributes to research on executive leadership, AI governance, and the design of AI-driven organizations. ## Joules-to-GDP: How Much Economic Value Do AI Systems Produce per Unit of Energy? DevFeed: [Joules-to-GDP: How Much Economic Value Do AI Systems Produce per Unit of Energy?]() Original publisher: [Read original article]() Author: Amirreza Zeinali, Jon Saad-Falcon, Avanika Narayan, Mersad Abbasi, Jonathan S. Hartley, John Hennessy, Christopher Re, Azalia Mirhoseini Published: 2026-10-23T04:00:00Z Content type: article Language: en Sources: [arXiv: Machine Learning]() Topics: [Artificial Intelligence](), [Machine learning](), [Sustainable Computing]() Tags: [accelerators](), [ai](), [ai-systems](), [cs-lg](), [econ-gn](), [energy](), [engineering](), [models](), [q-fin-ec](), [saving](), [software-development](), [value](), [work]() ### AI overview The study evaluates economic output per unit of end-to-end energy using two measures: net dollar value after execution cost per megajoule (NVJ) and expert time saved per megajoule (TSJ). Across 220 GDPval tasks in 44 occupations, higher value per task did not always mean better returns per joule: engineering tasks yielded about 10% more net value per task than software development, while software development had about 27% higher NVJ. GLM 5.2 had the highest average net value ($72.27 per task at 0.430 MJ); Gemma 4 31B produced 56% of that value with 23% of the energy, yielding about 2.5 times the NVJ and 2.8 times the TSJ. Hardware also mattered: Qwen3.6-27B's NVJ and TSJ varied substantially across four accelerators, with B200 leading both. Eight H100s saved 22% more completion time than one H100 at 3.24 times the energy, despite similar net value. ### Source excerpt arXiv:2610.30858v1 Announce Type: new Abstract: An AI system can produce more economically valuable work while delivering less value for the energy it consumes. We therefore ask: How much economic value does an AI system create per unit of energy consumed? We formalize economic value in two ways: (1) net dollar value, the output's dollar value minus AI-system execution cost, and (2) time saving, the expert's completion time minus the AI system's completion time. Dividing each by measured end-to-end energy yields Net Value per Joule (NVJ) and Time Saving per Joule (TSJ), reported in USD/MJ and hours/MJ. We study 220 public GDPval tasks across 44 occupations, serving five open-weight models locally and evaluating four accelerator types. First, the occupation groups where AI systems succeed most often, deliver the greatest dollar value, or save the most time do not necessarily provide the greatest economic return per joule. For example, among the tested tasks, engineering work delivers about 10% more net dollar value per task than software development. Yet software development delivers about 27% more net value per joule. Second, GLM 5.2 delivers the highest average net dollar value, at $72.27 per task using 0.430 MJ, equivalent in energy to about four hours of active laptop use. By contrast, Gemma 4 31B delivers 56% of that net dollar value using just 23% of the energy, achieving approximately 2.5 times the NVJ and 2.8 times the TSJ. Third, across four single-accelerator deployments, Qwen3.6-27B's NVJ varies by a factor of 4.32 and TSJ by 9.27, with B200 leading both. Additionally, one and eight H100s deliver similar net dollar value, but eight reduce completion time by 22% at 3.24 times the energy. Our framework makes output value and expert time saving under an energy budget explicit objectives for improving models, accelerators, and software, while offering a new basis for assessing AI's potential economic effect on occupational work. ## The Delegation Divide: An Open Index of AI-Agent Delegability across Occupations and Countries DevFeed: [The Delegation Divide: An Open Index of AI-Agent Delegability across Occupations and Countries]() Original publisher: [Read original article]() Author: Cristian Espinal Maya (Universidad Nacional de Colombia), Santiago Jim\\'enez Londo\\~no (Universidad EAFIT) Published: 2026-10-23T04:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [AI Agent]() Tags: [2026](), [adoption](), [agentic](), [agents](), [ai](), [ai-agent](), [ai-agents](), [arxiv](), [atlas](), [cs-ai](), [cs-cy](), [delegation](), [econ-gn](), [full](), [gap](), [generative](), [generative-ai](), [groups](), [index](), [jobs](), [language](), [language-models](), [open](), [q-fin-ec]() ### AI overview The Agent Delegability Index estimates that 35.2% of work time worldwide is at least partly delegable to AI agents in 2026, while only 2.5% can be fully delegated. It rates 18,796 O*NET task statements for delegation levels and barriers, maps them to occupations, and weights them using employment data from 146 countries. Estimated partial delegability ranges from 22.5% in low-income countries to 48.7% in high-income countries; the authors attribute this divide mainly to the physical demands of jobs common in poorer countries. ADI closely tracks existing AI exposure measures while adding delegation levels and barriers. Full delegation is rare, and the authors find no support for a wider adoption gap in middle-income countries. Model ratings show substantial agreement with human assessments, though the supplied text describes the abstract rather than the full paper. ### Source excerpt arXiv:2610.30752v1 Announce Type: cross Abstract: 35.2% of task time in world employment is at least partly delegable to AI agents in 2026, and only 2.5% can be handed over in full. Existing measures score exposure to generative AI, mostly for the United States. We build the Agent Delegability Index (ADI): two language models rate all 18,796 O*NET 30.2 task statements on a frozen, preregistered rubric of delegation levels and barriers; tasks are mapped to 433 ISCO-08 unit groups and weighted by ILOSTAT employment in 146 countries. Four findings. First, delegable employment rises with income, from 22.5% in low-income to 48.7% in high-income countries, a slope of 7.23 percentage points per log point of GDP per capita. This delegation divide is compositional: the physical barrier dominates the jobs poorer countries employ most. Second, at the occupation level ADI tracks exposure closely (r = 0.93 with Eloundou et al. and 0.89 with Felten et al.), above the preregistered bound of 0.80; ADI adds delegation levels and barriers on top of exposure, for the world, not a new ranking of occupations. Third, full delegation is rare and peaks for tasks that take a human about twelve minutes (11.8 minutes). Fourth, the preregistered adoption gap is not supported: across 101 countries, delegability and observed AI usage correlate at r = 0.86, and the gap between them is not wider in middle-income countries (p = 0.27). Model agreement is kappa = 0.77; a blind human rating of 300 tasks agrees with the model consensus at kappa = 0.85, and an agentic panel with web search at kappa = 0.87. We release the index, the atlas, the prompts and an MCP server at https://github.com/Cespial/delegation-divide. ## The Customer Is the Builder: How AI Coding Agents Are Rewriting the Economics of Software DevFeed: [The Customer Is the Builder: How AI Coding Agents Are Rewriting the Economics of Software]() Original publisher: [Read original article]() Author: Ripon Chandra Malo, Zimeng Yu, Tong Qiu Published: 2026-10-23T04:00:00Z Content type: article Language: en Sources: [arXiv: Software Engineering]() Topics: [agentic-coding](), [AI-assisted coding](), [Open Source]() Tags: [2023](), [about](), [abstract](), [agents](), [ai](), [ai-coding](), [ai-coding-agents](), [alternatives](), [application](), [arxiv](), [benchmark](), [build](), [builder](), [but](), [can](), [coding](), [coding-agents](), [cs-cy](), [cs-se](), [customer](), [econ-gn](), [economics](), [q-fin-ec](), [software]() ### AI overview AI coding agents may let organizations build some software functionality instead of buying it, but evidence of a broad market shift is mixed. Since 2023, leading agents' benchmark task length at 50% success has doubled about every 129 days. A search found 575 potential open-source alternatives to selected commercial products, and the Craft suite produced 12 application repositories in 12 days; functional matches need validation, and Craft versions have unverified completeness and reliability. While purchased software gained investment share and tracked commercial installations grew, 32% of surveyed respondents said their organization had forgone at least one purchase because it could build the functionality with coding agents. Those self-reports do not establish displaced spending or tool reliability. Benchmark projections extrapolate capability, not customer switching. ### Source excerpt arXiv:2610.30307v1 Announce Type: new Abstract: AI coding agents could change whether organizations buy software or build the functionality they need. We focus on organizations and professional buyers whose domain knowledge can help them specify and evaluate tools for their work. We define the software substitution frontier as the functions customers could build and maintain at an acceptable standard for less than buying them. Using public data, we compare growing capacity to build with indicators of market change. Since 2023, the length of benchmark tasks that leading agents can complete at 50 percent success, measured in human working time, has doubled about every 129 days. A broad search identified 575 potential open-source alternatives to selected commercial products, although some functional matches still require independent validation. The Craft suite produced 12 application repositories over 12.0 days of public development, but these remain early versions with unverified completeness and reliability. Market evidence is mixed: purchased software gained US investment share and tracked commercial installations grew. Yet 32 percent of respondents to an organizational survey reported forgoing at least one software purchase because their organization could build the functionality with coding agents. These self-reports do not establish the spending displaced or the reliability of the resulting tools. The gap motivates a warning: customers' capacity to build could affect individual buying decisions before a broad purchasing shift becomes visible. We examine this possibility using benchmark projections and observable market tests. The projections are an extrapolation, not a forecast of customer switching. ## Motivation and Contribution in Open Source Software: A Meta-Analytic Investigation DevFeed: [Motivation and Contribution in Open Source Software: A Meta-Analytic Investigation]() Original publisher: [Read original article]() Author: Parisa Omidmand, Saeed Nosratabadi Published: 2026-10-22T04:00:00Z Content type: article Language: en Sources: [arXiv: Software Engineering]() Topics: [Open Source](), [intrinsic motivation]() Tags: [contribution](), [cs-hc](), [cs-se](), [econ-gn](), [meta](), [motivation](), [open-source](), [q-fin-ec]() ### AI overview A meta-analysis of 41 studies finds that motivations relate more strongly to contributors' intention to continue than to their actual effort. The main analysis pools 178 correlations covering eight antecedents across 29 independent samples (N=9,632): seven antecedents are positively related to effort, all eight to continuance intention, and each relates at least as strongly to intention. Intrinsic motivation is not more strongly related to effort than extrinsic motivation, though its intention-effort gap is larger. Reputation relates to effort mainly on platforms with reputation systems; the analysis finds no evidence that monetary rewards crowd out enjoyment. ### Source excerpt arXiv:2610.29199v1 Announce Type: cross Abstract: Research on why people contribute to open source software (OSS) and related open collaboration platforms spans more than two decades, but its findings conflict, and studies often treat the work people contribute and their intention to keep contributing as interchangeable. Drawing on self-determination theory and motivation crowding theory, we meta-analyze 41 studies. The main analysis pools 178 correlations of eight antecedents with contribution effort and continuance intention from 29 independent samples (N = 9,632). Seven antecedents relate positively to effort, all eight relate positively to intention, and each relates at least as strongly to intention as to effort. Contrary to a common assumption, intrinsic motivation is not more strongly related to effort than extrinsic motivation, but its gap between intention and effort is larger. Reputation relates to effort mainly on platforms with a reputation system, and we find no sign that monetary rewards crowd out enjoyment. The findings inform how platforms can turn contributors' intentions into work. ## JaxMARL-BC: Multi-Agent Reinforcement Learning for Macroeconomic Modelling, in Seconds DevFeed: [JaxMARL-BC: Multi-Agent Reinforcement Learning for Macroeconomic Modelling, in Seconds]() Original publisher: [Read original article]() Author: Daniele Montagnani, Andrea Coletta, Antonietta Mira, Aldo Glielmo Published: 2026-10-22T04:00:00Z Content type: news Language: en Sources: [arXiv: Machine Learning]() Topics: [Reinforcement learning](), [Multi Agent Systems](), [Model Development]() Tags: [cs-ai](), [cs-lg](), [cs-ma](), [econ-gn](), [jax](), [modelling](), [multi-agent](), [open-source](), [q-fin-ec](), [reinforcement-learning](), [scaling](), [simulator](), [training]() ### AI overview JaxMARL-BC is a JAX-accelerated simulator designed to reduce the long training times that have limited multi-agent reinforcement learning in macroeconomic models. The authors report training 75-724 times faster than the reference MARL-BC implementation and scaling to 200,000 independent learning households. They check correctness against known DSGE limits and describe scaling with a simple cost law. In a proof-of-concept calibration of a model with heterogeneous returns on capital, a two-hour experiment replaced work estimated to take more than a month with the prior implementation. The abstract announces an open-source release; it does not provide implementation details or broader evaluation results. ### Source excerpt arXiv:2610.28907v1 Announce Type: cross Abstract: Multi-agent reinforcement learning (MARL) offers a principled way of populating simulated economies with many heterogeneous decision makers. Its adoption, pioneered in finance, has nonetheless been slow in macroeconomics because of two main limitations: the lack of a shared frame of reference connecting MARL systems to dynamic stochastic general equilibrium (DSGE) models the profession already trusts, and the very long training times. The first limitation was recently addressed by the Multi-Agent Reinforcement Learning Business Cycle (MARL-BC), which places deep RL households in a real business cycle economy and reduces to classic DSGE models in specific limits. Here we address the second limitation. We present JaxMARL-BC, a JAX-accelerated simulator that trains MARL-BC economies in seconds (between 75x and 724x faster than the reference implementation), and scales to 200,000 independent learning households. We verify its correctness against the known DSGE limits, characterise its scaling through a simple cost law, and demonstrate its value by a proof-of-concept calibration of an extended model with heterogeneous returns on capital, in a two-hour experiment that would previously have taken over a month. We release JaxMARL-BC as open source.