Knowledge that Transforms

As artificial intelligence (AI) tools have become pervasive in business applications, so too have interactions between AI and humans in business processes and decision-making. A growing area of research has focused on human decision and task delegation to AI assistants. Simultaneously, extensive research on algorithm aversion—humans’ resistance to algorithm-based decision tools—has demonstrated potential barriers and issues with AI applications in business. In this paper, we test a simple strategy for mitigating algorithm aversion in the context of AI task delegation. We show that simply changing the framing of decision tasks can allay algorithm aversion. Through multiple studies, we found that participants exhibited a strong preference for human assistance over AI assistance when they were rewarded for task performance (i.e., money was gained for good performance), even when the AI had been shown to outperform the human assistant on the task. Alternatively, when we reframed the task such that the participant experienced losses for poor performance (i.e., money was taken from their endowment for poor performance), the bias for preferring human assistance was removed. Under loss framing, participants delegated the decision task to human and AI assistants at similar rates. We demonstrate this finding across tasks at differing levels of complexity and at different incentive sizes. We also provide evidence that loss framing increases situational awareness, which drives the observed effects. Our results offer useful insights on reducing algorithm aversion that extend the literature and provide actionable suggestions for practitioners and managers. This paper was accepted by Dongjun Wu, Special Issue on the Human-Algorithm Connection. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.05585 .

Performance ranking can trigger multiple social incentives for workers. On one hand, it offers status rewards to induce them to increase effort. On the other, better-ranked workers may reduce effort to conform to coworkers’ productivity in fear of social retribution. This paper uses a field experiment in a sweater factory to disentangle the incentives underlying performance ranks. Treated workers receive ranks either privately or publicly. I find that private ranks do not have any effect on average but that public ranks reduce worker productivity. Additional evidence confirms that productivity declines because of workers’ social concerns and their desire to conform to the productivity of their friends. Cooperation between workers decreases too but with limited effect on productivity. The paper illustrates how inducing worker competition may be counterproductive for firms. This paper was accepted by Marie Claire Villeval, behavioral economics and decision analysis. Funding: Financial support from the Private Enterprise Development in Low-Income Countries, Centre for Economic Policy Research [PEDL ERG 4362] and the DFG through CRC TRR 190 [Project 280092119] are gratefully acknowledged. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01381 .

We develop an end-to-end, scalable machine learning framework for designing bundle recommendations in a high-dimensional choice setting. We leverage historical purchases and consideration sets determined from clickstream data to generate dense representations (embeddings) of products. We impose minimal structure on these embeddings and develop heuristics for complementarity and substitutability among products. Subsequently, we use the heuristics to create multiple bundle recommendations for each of 4,500 focal products and test their performance using a field experiment with a large retailer. We use the experimental data to optimize the recommendation design policy with offline policy learning. Our optimized policy is robust across product categories, generalizes well to the retailer’s entire assortment, and provides an expected improvement of 35% ([Formula: see text] per 100 visits) in revenue from bundle recommendations over the baseline policy. This paper was accepted by Hemant Bhargava, information systems. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.01864 .

We document that traditional energy firms are key innovators in the United States’ green patent landscape. These firms produce more, and significantly higher-quality, green innovation. In many green technology spaces, they appear to be influential first movers and to produce ongoing foundational aspects of innovation and commercialization on which other alternative energy producers build. They additionally invest significantly in labor and capital to complement these green innovations. These traditional energy firms, however, receive significantly lower environmental, social, and governance (ESG) scores and fund flows and are not rewarded for incremental green innovation. This behavior is consistent with a competitive response by traditional energy firms to preempt obsolescence of current technology by investing in future replacement technologies. This paper was accepted by Bo Becker, finance. Funding: Funding was provided by the National Science Foundation [Grant SciSIP 1535813] and the Fordham University Gabelli School of Business—PVH Corp. Global Thought Leadership Grant on Corporate Social Responsibility. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.09124 .

There is growing agreement among researchers and developers that in certain machine-learning (ML) tasks, it may be advantageous to keep a “human in the loop” rather than rely on fully autonomous systems. Continual human involvement can mitigate machine bias and performance deterioration while enabling humans to continue learning from insights derived by ML. Yet a microlevel theory that effectively facilitates joint and continual learning in both humans and machines is still lacking. To address this need, we adopt a design science approach and build on theories of human reciprocal learning to develop an abstract configuration for reciprocal human-ML (RHML) in the context of text message classification. This configuration supports learning cycles between humans and machines who repeatedly exchange feedback regarding a classification task and adjust their knowledge representations accordingly. Our configuration is instantiated in Fusion, a novel technology artifact. Fusion is developed iteratively in two case studies of cybersecurity forums (drug trafficking and hacker attacks), in which domain experts and ML models jointly learn to classify textual messages. In the final stage, we conducted two experiments of the RHML configuration to gauge both human and machine learning processes over eight learning cycles. Generalizing our insights, we provide formal design principles for the development of systems to support RHML. This paper was accepted by D. J. Wu, Special Issue on the Human-Algorithm Connection. Funding: This work was supported by the Israel’s Ministry of Defence [Grant R4441197567] and the Israel’s Ministry of Science and Technology [Grant 207076]. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2022.03518 .

In decentralized governance, quadratic voting (QV)—where the cost of acquiring voting power is convex—optimally aggregates voter preferences, outperforming simpler linear voting (LV) mechanisms when voters have complete information. But what if they do not? We show that uncertainty not only breaks QV optimality but can also cause it to underperform LV. Intuitively, this is because cost convexity can disincentivize better-informed voters from adequately conveying their private information. The optimal mechanism varies with the distribution of stakes and information among voters, implying that QV’s known advantages in preference aggregation do not readily extend to common-value information aggregation settings. This paper was accepted by Will Cong for the Special Issue on the Digital Finance. Supplemental Material: The online appendices are available at https://doi.org/10.1287/mnsc.2024.08469 .

This paper considers introducing asymmetric privacy in the design of central bank digital currencies (CBDC) and digital currencies more generally to preserve the privacy of money spent while keeping the benefits of digital records for money received. It is shown that this feature would help minimize real distortions between consumers, firms, and financiers while enabling tax optimization and better access to external financing. Protecting the privacy of consumers is desirable from a welfare and efficiency standpoint as long as there exist noticeable privacy concerns. Implementing asymmetric privacy is technologically feasible, using, for instance, zero-knowledge proofs or other privacy tools. This paper has been accepted by Lin William Cong for the Virtual Special Issue on Digital Finance. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2024.06830 .

This study shows that competition drives corporate innovation under intense environmental regulatory pressure. Using the nonattainment status of U.S. counties as an exogenous variation in regulation, we find that competition spurs green innovation as firms respond to stricter policies. Firms are particularly motivated to innovate in clean technology when operating in pollution-intensive industries, facing high relocation costs, and possessing a strong history of innovation. Regulation-driven green innovation allows firms to differentiate their products, enhance their environmental, social, and governance (ESG) reputation, and attract more corporate customers, leading to higher sales growth, increased market share, and improved profitability, although not necessarily higher valuation. Stricter regulations in competitive environments not only curb pollution but also serve as a catalyst for sustainable long-term innovation. These findings emphasize the vital role of environmental regulations in promoting sustainable practices and operational benefits, underscoring the importance of well-designed policies to drive long-term economic and environmental progress. This paper was accepted by Bo Becker, finance. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.00600 .

Using confidential information on banks’ portfolios, we show that banks that emphasize the sustainability of their lending policies in their disclosures do not exhibit a reduced environmental impact and, if anything, they extend a higher volume of credit to brown borrowers, without charging higher interest rates, shortening debt maturity, or requiring more collateral. These results cannot be attributed to the financing of borrowers’ transition toward greener technologies. Examining the mechanisms behind the strategic disclosure choices reveals that banks extend credit to existing brown borrowers, especially those who are financially underperforming. This paper was accepted by Caroline Flammer, sustainability. Funding: M. Giannetti acknowledges financial support from the Swedish House of Finance, the Nasdaq Nordic Foundation, the Karl-Adam Bonnier Foundation, and the Jan Wallander and Tom Hedelius Foundation. M. Jasova acknowledges financial support from the Barnard College Presidential Research Award. M. Loumioti acknowledges financial support from the University of Texas at Dallas. The opinions expressed herein are those of the authors and do not necessarily reflect those of the ECB or the Eurosystem. All errors are the authors’ own. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.07420 .

Managerial decision makers are increasingly supported by advanced data analytics and other artificial intelligence (AI)-based technologies, but they are often found to be hesitant to follow the algorithmic advice. We examine how compensation contract design and framing of an AI algorithm influence decision makers’ reliance on algorithmic advice and performance in a price estimation task. Based on a large sample of almost 1,500 participants, we find that compared with a fixed compensation, both compensation contracts based on individual performance and tournament contracts lead to an increase in effort duration and to more reliance on algorithmic advice. We further find that using an AI algorithm that is framed as also incorporating human expertise has positive effects on advice utilization, especially for decision makers with fixed pay contracts. By showing how widely used control practices, such as incentives and task framing, influence the interaction of human decision makers with AI algorithms, our findings have direct implications for managerial practice. This paper was accepted by David Simchi-Levi, Special Issue on the Human-Algorithm Connection. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.02777 .