ISOM Research Workshop

Thursday, February 13, 2025

• 6:30 pm: Workshop kick-off dinner at Mildred’s Big City Food,
  • 3445 W. University Avenue
    352-371-1711

Friday, February 14, 2025

• 8:00 am: Continental breakfast, Bryan Hall Room 232
• 8:30 am: Saby Mitra, Dean, Warrington College of Business
• 8:50 am: Workshop introduction, Emre Demirezen, Workshop Co-organizer
• 9:00 am
  • Environmental Impacts and Operational Practices of Facilities in Minority Communities
    Ravi Subramanian, Georgia Institute of Technology
• 9:45 am
  • AI agents for non-profit oOrganizations
    Pallab Sanyal, George Mason University
• 10:30 am: Coffee break
• 11:00 am
  • Information Design in Ghost Kitchens? How Human-Tech Overrides Affect Order Fulfillment
    Neha Sharma, University of Pennsylvania
• 11:45 am
  • AI Enforcement: Examining the Impact of AI on Judicial Fairness and Public Safety
    Yi-Jen (Ian) Ho, Tulane University
• 12:30 pm: Lunch
• 2:00 pm
  • Getting to the Green: Should a Profit-Maximizing Firm Buy Carbon Offsets or Invite Consumers to Buy Them?
    Gokce Esenduran, Purdue University
• 2:45 pm
  • Of the first five US states with food waste bans, Massachusetts alone has reduced landfill waste
    Ioannis Stamatopoulos, University of Texas-Austin
• 3:30 pm: Coffee break
• 4:00 pm
  • Survival Analysis for Managing Animal Rescue
    Elena Katok/Qiuxia (Katalia) Chen, University of Texas at Dallas
• 6:30 pm
  • Workshop Reception at Spurrier’s Grid Iron Grill
    4860 Steve Spurrier Way (Celebration Pointe)
    352-500-4422

Saturday, February 15, 2025

• 8:30 am: Continental breakfast, Bryan Hall Room 232
• 9:00 am
  • The Double-Edged Roles of Generative AI in the Creative Process: Experiments on Design Work
    Gang Wang, University of Delaware
• 9:45 am
  • When Puffery Advertising Meets Generative AI: Evidence from Two Field Studies
    Ling Xue, University of Georgia
• 10:30 am: Coffee break
• 11:00 am
  • Is Love Serendipitous? Exploring the Impact of a Serendipity-Oriented Recommender System in Online Dating Through a Randomized Field Experiment
    Yumei He, Tulane University
• 11:45 am
  • Concluding remarks, Asoo J. Vakharia, Workshop Co-organizer
• 12:00 pm: Boxed lunch

Ravi Subramaniam, Georgia Institute of Technology

Environmental Justice (EJ) encapsulates the idea of fairness in the protection of individuals and communities from environmental hazards, regardless of economic or social background. EJ is relevant to the practice of operations management because of inequities that may result from socio-geographically disparate operational practices. Our work studies an important research gap by investigating facility-level heterogeneity in the environmental impacts imposed on communities and in the use of operational practices involved in the generation, avoidance, and mitigation of these impacts, thus advancing the emphasis on EJ from an aggregative geographic level to the facility level, where operational decisions are made.

We draw on comprehensive chemical release and environmental risk data from the US EPA’s Toxics Release Inventory and Risk-Screening Environmental Indicators Model, community-level data from the US Census Bureau’s American Community Survey, and facility-level data from the National Establishment Time Series. We employ matching and instrumented panel data methods in our analysis. We find consistent evidence of greater environmental impacts by facilities operating in minority communities, which persists across a variety of robustness checks. However, we do not find significantly different environmental impact avoidance or mitigation levels by facilities in minority communities, compared to similar facilities in non-minority communities. While seemingly encouraging, such an approach is deficient from the standpoint of EJ because of the resulting persistence of contemporaneous disparities. For additional managerial and policy insights, we examine operational, organizational, and institutional drivers of environmental impact disparities between facilities operating in minority and non-minority communities.

Environmental Justice (EJ) encapsulates the idea of fairness in the protection of individuals and communities from environmental hazards, regardless of economic or social background. EJ is relevant to the practice of operations management because of inequities that may result from socio-geographically disparate operational practices. Our work studies an important research gap by investigating facility-level heterogeneity in the environmental impacts imposed on communities and in the use of operational practices involved in the generation, avoidance, and mitigation of these impacts, thus advancing the emphasis on EJ from an aggregative geographic level to the facility level, where operational decisions are made.

We draw on comprehensive chemical release and environmental risk data from the US EPA’s Toxics Release Inventory and Risk-Screening Environmental Indicators Model, community-level data from the US Census Bureau’s American Community Survey, and facility-level data from the National Establishment Time Series. We employ matching and instrumented panel data methods in our analysis. We find consistent evidence of greater environmental impacts by facilities operating in minority communities, which persists across a variety of robustness checks. However, we do not find significantly different environmental impact avoidance or mitigation levels by facilities in minority communities, compared to similar facilities in non-minority communities. While seemingly encouraging, such an approach is deficient from the standpoint of EJ because of the resulting persistence of contemporaneous disparities. For additional managerial and policy insights, we examine operational, organizational, and institutional drivers of environmental impact disparities between facilities operating in minority and non-minority communities.

Pallab Sanyal, George Mason University
(joint work with Scott Schanke and Siddharth Bhattacharya)

Conversational AI agents, such as chatbots and virtual assistants, are increasingly becoming integral to businesses across various industries, revolutionizing customer service, sales, and operations. These intelligent systems leverage natural language processing (NLP) and machine learning to interact with users, automate repetitive tasks, and provide personalized experiences. The focus of our study is on non-profit and charitable organizations, with the goal of helping them scale their operations and alleviate the challenges associated with limited resources. To that end, the study investigates the efficacy of AI chatbots designed for nonprofits, diverging from traditional business-centric AI-based conversational agents. By collaborating with a Minneapolis-based non-profit organization, the research examines the role of emotional versus informational appeals and anthropomorphism in chatbot interactions within a randomized field experiment. We find that the number of emotional appeals in the interaction and increased anthropomorphism both negatively impact conversion. We also find that increasing emotional appeals evoke a highly emotional response, leading to reactance which also leads to lower conversion. These findings suggest that traditional means of persuasion, social cues, and emotional appeals might not fit charitable donation contexts and may even harm donor interactions. The study provides valuable insights for chatbot design tailored to non-profit organizations and emphasizes crafting appropriate AI-driven messaging for effective outreach.

Conversational AI agents, such as chatbots and virtual assistants, are increasingly becoming integral to businesses across various industries, revolutionizing customer service, sales, and operations. These intelligent systems leverage natural language processing (NLP) and machine learning to interact with users, automate repetitive tasks, and provide personalized experiences. The focus of our study is on non-profit and charitable organizations, with the goal of helping them scale their operations and alleviate the challenges associated with limited resources. To that end, the study investigates the efficacy of AI chatbots designed for nonprofits, diverging from traditional business-centric AI-based conversational agents. By collaborating with a Minneapolis-based non-profit organization, the research examines the role of emotional versus informational appeals and anthropomorphism in chatbot interactions within a randomized field experiment. We find that the number of emotional appeals in the interaction and increased anthropomorphism both negatively impact conversion. We also find that increasing emotional appeals evoke a highly emotional response, leading to reactance which also leads to lower conversion. These findings suggest that traditional means of persuasion, social cues, and emotional appeals might not fit charitable donation contexts and may even harm donor interactions. The study provides valuable insights for chatbot design tailored to non-profit organizations and emphasizes crafting appropriate AI-driven messaging for effective outreach.

Neha Sharma, University of Pennsylvania

Cloud or ghost kitchens operate without store fronts and serve take-out and home-delivery customers. Recently, multi-brand cloud kitchens that mimic food courts have emerged, allowing customers to order from multiple brands in a single order. Coordinating such orders poses unique challenges, making the fulfillment process critical. We investigate the interplay between brand chefs and the platform’s intended fulfillment process. Specifically, we examine the extent of information the platform should share with chefs for orders requiring coordination between brands, the consequences of chefs independently prioritizing tasks, and whether the platform should prioritize orders that demand coordination. We collaborate with a large Indian multi-brand cloud kitchen platform and collect micro-level data on 6.6 million orders spanning 16 months. Our empirical analysis revealed: (1) Orders involving multiple brands account for only 0.01% to 2% of total orders but experience fulfillment times that are 50\% longer than single-brand orders. (2) To boost sales, these brands were collaborating by including dishes from other brands in their menus. Therefore, 30-40% of orders that appear to be single-brand actually consist of dishes from multiple brands, which surprisingly have similar fulfillment times to single-brand orders. Field visits revealed this is due to chefs having more information about such orders and, subsequently, overriding the platform’s chosen sequence.

Using a fork-join queuing network model and the Simulated Method of Moments (SMM) approach, we evaluate fulfillment times and quality metrics under various alternate policies. Sharing more information about multi-brand orders with chefs enables the platform to achieve Pareto improvement wherein it can not only reduce the order fulfillment time (by 0.10-4.7%) and improve quality of temperature-sensitive orders (by 0.01 – 2.44%) over the current policy across different order classes but also create more equity in outcomes across order classes.

Cloud or ghost kitchens operate without store fronts and serve take-out and home-delivery customers. Recently, multi-brand cloud kitchens that mimic food courts have emerged, allowing customers to order from multiple brands in a single order. Coordinating such orders poses unique challenges, making the fulfillment process critical. We investigate the interplay between brand chefs and the platform’s intended fulfillment process. Specifically, we examine the extent of information the platform should share with chefs for orders requiring coordination between brands, the consequences of chefs independently prioritizing tasks, and whether the platform should prioritize orders that demand coordination. We collaborate with a large Indian multi-brand cloud kitchen platform and collect micro-level data on 6.6 million orders spanning 16 months. Our empirical analysis revealed: (1) Orders involving multiple brands account for only 0.01% to 2% of total orders but experience fulfillment times that are 50\% longer than single-brand orders. (2) To boost sales, these brands were collaborating by including dishes from other brands in their menus. Therefore, 30-40% of orders that appear to be single-brand actually consist of dishes from multiple brands, which surprisingly have similar fulfillment times to single-brand orders. Field visits revealed this is due to chefs having more information about such orders and, subsequently, overriding the platform’s chosen sequence.

Using a fork-join queuing network model and the Simulated Method of Moments (SMM) approach, we evaluate fulfillment times and quality metrics under various alternate policies. Sharing more information about multi-brand orders with chefs enables the platform to achieve Pareto improvement wherein it can not only reduce the order fulfillment time (by 0.10-4.7%) and improve quality of temperature-sensitive orders (by 0.01 – 2.44%) over the current policy across different order classes but also create more equity in outcomes across order classes.

Yi-Jen (Ian) Ho, Tulane University
(joint work with Wael Jabr and Yifan Zhang)

The judicial system in the United States faces the challenge of prison overcrowding and record-high incarceration costs. To alleviate these challenges, artificial intelligence (AI) is increasingly adopted to assess offenders’ risk of recidivism and recommend alternative punishments for low-risk offenders. Although AI provides objective recommendations, its impact on judges’ decision-making, judicial fairness toward offenders, and public safety remains debatable. To provide empirical insights into these significant debates, we apply regression discontinuity identification to a unique dataset comprising 27,357 sentencing cases in Virginia from 2013 to 2022. We show that adopting an AI instrument significantly affects judges’ decisions by increasing the probability of offering alternative punishments, lowering the probability of incarceration, and shortening sentence terms. More importantly, we find that the AI’s recommendations impact judicial fairness in two opposite directions. On the one hand, while judges are typically more lenient with female offenders than males, AI helps alleviate such gender-based disparity. On the other hand, as judges maintain fair sentencing without the presence of the recommendations, we find evidence of a racial bias favoring White offenders over Black ones with the recommendations. We further analyze the societal impact of judges’ decisions on public safety regarding offenders’ recidivism. We confirm that judges’ discretion regarding AI’s recommendations on offenders’ genders is welcome, but their favor toward White offenders hurts public safety. Accordingly, we provide actionable implications for judges, policymakers, and the public to promote judicial fairness with AI support.

The judicial system in the United States faces the challenge of prison overcrowding and record-high incarceration costs. To alleviate these challenges, artificial intelligence (AI) is increasingly adopted to assess offenders’ risk of recidivism and recommend alternative punishments for low-risk offenders. Although AI provides objective recommendations, its impact on judges’ decision-making, judicial fairness toward offenders, and public safety remains debatable. To provide empirical insights into these significant debates, we apply regression discontinuity identification to a unique dataset comprising 27,357 sentencing cases in Virginia from 2013 to 2022. We show that adopting an AI instrument significantly affects judges’ decisions by increasing the probability of offering alternative punishments, lowering the probability of incarceration, and shortening sentence terms. More importantly, we find that the AI’s recommendations impact judicial fairness in two opposite directions. On the one hand, while judges are typically more lenient with female offenders than males, AI helps alleviate such gender-based disparity. On the other hand, as judges maintain fair sentencing without the presence of the recommendations, we find evidence of a racial bias favoring White offenders over Black ones with the recommendations. We further analyze the societal impact of judges’ decisions on public safety regarding offenders’ recidivism. We confirm that judges’ discretion regarding AI’s recommendations on offenders’ genders is welcome, but their favor toward White offenders hurts public safety. Accordingly, we provide actionable implications for judges, policymakers, and the public to promote judicial fairness with AI support.

Gokce Esenduran, Purdue University
(joint work with H. Sebastian Heese and Gilvan Souza)

Many firms, such as Delta Airlines, Patagonia, Google, and Apple, have publicly stated their goal of becoming carbon neutral at some point in the future. These firms are pursuing multiple emissions reductions initiatives within their value chains, such as the use of renewable energy, zero-emissions vehicles, and low-carbon materials and supplies. But despite these efforts, some residual emissions cannot be further reduced, necessitating the purchase of carbon offsets, which increase firms’ costs. While environmentally conscious consumers may be willing to pay a higher price for a low-emissions product or service, a significant segment of climate change–disengaged consumers are not willing to do so, as demonstrated by multiple studies. In this paper, we identify when it is optimal for a profit-maximizing firm to offer its consumers carbon-reduction offsets for purchase with the product, in addition to potentially purchasing offsets at the firm level.

Through a stylized analytical model, we show that it is never optimal for firms to both buy offsets at the firm level and offer them to consumers for purchase at the same time. Rather, when the offset cost is low enough, the firm buys enough offsets to compensate for all of its emissions at the firm level, resulting in a higher overall product price; when the offset cost is in the middle range, the firm offers offsets to consumers for purchase at an offset price lower than offset cost, ensuring that green consumers buy offsets; and when the offset price is high, offsets cannot be optimally used in any way. The thresholds depend on the green-segment size and disutility from emissions (at the consumer-consumption and firm levels), and on the product’s own carbon footprint.

Providing offsets for purchase to consumers allows consumers to self-select into their preferred product/price bundle, thus providing the firm with a market-segmentation and price-discrimination mechanism that increases profits and reduces the firm’s carbon footprint.

Many firms, such as Delta Airlines, Patagonia, Google, and Apple, have publicly stated their goal of becoming carbon neutral at some point in the future. These firms are pursuing multiple emissions reductions initiatives within their value chains, such as the use of renewable energy, zero-emissions vehicles, and low-carbon materials and supplies. But despite these efforts, some residual emissions cannot be further reduced, necessitating the purchase of carbon offsets, which increase firms’ costs. While environmentally conscious consumers may be willing to pay a higher price for a low-emissions product or service, a significant segment of climate change–disengaged consumers are not willing to do so, as demonstrated by multiple studies. In this paper, we identify when it is optimal for a profit-maximizing firm to offer its consumers carbon-reduction offsets for purchase with the product, in addition to potentially purchasing offsets at the firm level.

Through a stylized analytical model, we show that it is never optimal for firms to both buy offsets at the firm level and offer them to consumers for purchase at the same time. Rather, when the offset cost is low enough, the firm buys enough offsets to compensate for all of its emissions at the firm level, resulting in a higher overall product price; when the offset cost is in the middle range, the firm offers offsets to consumers for purchase at an offset price lower than offset cost, ensuring that green consumers buy offsets; and when the offset price is high, offsets cannot be optimally used in any way. The thresholds depend on the green-segment size and disutility from emissions (at the consumer-consumption and firm levels), and on the product’s own carbon footprint.

Providing offsets for purchase to consumers allows consumers to self-select into their preferred product/price bundle, thus providing the firm with a market-segmentation and price-discrimination mechanism that increases profits and reduces the firm’s carbon footprint.

Ioannis (Yannis) Stamatopoulos, University of Texas-Austin
(joint work with Fiorentia Zoi Anglou and Robert Evan Sanders)

Diverting food waste from landfills is crucial to reduce emissions and meet Paris Agreement targets. Between 2014 and 2024, nine US states banned commercial waste generators—such as grocery chains—from landfilling food waste, expecting a 10 to 15% waste reduction. However, no evaluation of these bans exists. We compile a comprehensive waste dataset covering 36 US states between 1996 and 2019 to evaluate the first five implemented state-level bans. Contrary to policy-makers’ expectations, we can reject aggregate waste reductions higher than 3.2%, and we cannot reject a zero-null aggregate effect. Moreover, we cannot reject a zero-null effect for any other state except Massachusetts, which gradually achieved a 13.2% reduction. Our findings reveal the need to reassess food waste bans using Massachusetts as a benchmark for success.

Diverting food waste from landfills is crucial to reduce emissions and meet Paris Agreement targets. Between 2014 and 2024, nine US states banned commercial waste generators—such as grocery chains—from landfilling food waste, expecting a 10 to 15% waste reduction. However, no evaluation of these bans exists. We compile a comprehensive waste dataset covering 36 US states between 1996 and 2019 to evaluate the first five implemented state-level bans. Contrary to policy-makers’ expectations, we can reject aggregate waste reductions higher than 3.2%, and we cannot reject a zero-null aggregate effect. Moreover, we cannot reject a zero-null effect for any other state except Massachusetts, which gradually achieved a 13.2% reduction. Our findings reveal the need to reassess food waste bans using Massachusetts as a benchmark for success.

Elena Katok and Qiuxia (Katalia) Chen, University of Texas-Dallas

Municipal shelters must euthanize numerous animals (primarily dogs) every day due to a shortage of space. The choice of which dogs to euthanize is based on two policy decisions: the length of time the dog is given at the shelter, and the degree to which the shelter capacity can be expanded. The more the shelter is over capacity, the more stressful it is for the dogs who are there, due to noise, overcrowding, and a lower quality of care, therefore, shelter managers have to consider the trade-off between the higher probability of a dog being adopted when it stays longer, and the lower probability of a dog being adopted due to over-crowded conditions. Using data from a large municipal shelter in Dallas, we derive a method for determining the optimal policies using survival analysis and queuing model. This analysis allows us to qualitatively estimate the number of lives saved compared to the current practice.

Municipal shelters must euthanize numerous animals (primarily dogs) every day due to a shortage of space. The choice of which dogs to euthanize is based on two policy decisions: the length of time the dog is given at the shelter, and the degree to which the shelter capacity can be expanded. The more the shelter is over capacity, the more stressful it is for the dogs who are there, due to noise, overcrowding, and a lower quality of care, therefore, shelter managers have to consider the trade-off between the higher probability of a dog being adopted when it stays longer, and the lower probability of a dog being adopted due to over-crowded conditions. Using data from a large municipal shelter in Dallas, we derive a method for determining the optimal policies using survival analysis and queuing model. This analysis allows us to qualitatively estimate the number of lives saved compared to the current practice.

Ling Xue, University of Georgia
(joint work with Yueyue Zhang, Cheng Zhang, Fu Zhang)

With puffery advertising being a prevalent practice, there is concern about the implications of this tool for information presentation and manipulation in advertising. This study investigates the role of generative artificial intelligence (GAI) in addressing puffery advertising. Using two field studies, we specifically examine how GPT, a major GAI tool, can be used by a demand-side platform (DSP) to correct puffery content created by advertisers, so as to comply with media platforms’ restrictions for launching targeted advertisements. In the first study, the DSP uses GPT to correct advertisements flagged for puffery by media platforms. Due to varying puffery tolerance policies, the same advertisement was revised for some media platforms but not others. Analyzing over 295,000 real advertising exposures, we found that by using legitimate content to replace puffery, GPT revision can significantly increase the probability of advertisement click by 16%. We further conduct a randomized field experiment, in which we use prompt engineering to guide the GPT revision of each individual linguistic and emotional feature of the advertisement. The results confirm the mechanisms through which the GPT revision enhances the advertisement performance. They also reveal that the improvement in linguistic readability is the most effective in turning puffery advertisements into attractive, legitimate ones. The study generates important implications on how GAI can be used to effectively address puffery advertising and increase marketing performance. It also illustrates that puffery advertising may not always be as luring as it may appear. The tackling of puffery advertising by GAI can not only resolve ethical concerns in advertising but also enhance advertising performance.

With puffery advertising being a prevalent practice, there is concern about the implications of this tool for information presentation and manipulation in advertising. This study investigates the role of generative artificial intelligence (GAI) in addressing puffery advertising. Using two field studies, we specifically examine how GPT, a major GAI tool, can be used by a demand-side platform (DSP) to correct puffery content created by advertisers, so as to comply with media platforms’ restrictions for launching targeted advertisements. In the first study, the DSP uses GPT to correct advertisements flagged for puffery by media platforms. Due to varying puffery tolerance policies, the same advertisement was revised for some media platforms but not others. Analyzing over 295,000 real advertising exposures, we found that by using legitimate content to replace puffery, GPT revision can significantly increase the probability of advertisement click by 16%. We further conduct a randomized field experiment, in which we use prompt engineering to guide the GPT revision of each individual linguistic and emotional feature of the advertisement. The results confirm the mechanisms through which the GPT revision enhances the advertisement performance. They also reveal that the improvement in linguistic readability is the most effective in turning puffery advertisements into attractive, legitimate ones. The study generates important implications on how GAI can be used to effectively address puffery advertising and increase marketing performance. It also illustrates that puffery advertising may not always be as luring as it may appear. The tackling of puffery advertising by GAI can not only resolve ethical concerns in advertising but also enhance advertising performance.

Gang Wang, University of Delaware

Generative Artificial Intelligence (GenAI) is reshaping creative domains by offering novel and complex content solutions with transformative potential. However, the existing GenAI literature largely examines creativity as an end product, overlooking the intricate dynamics of the human-GenAI co-creative process. Addressing this gap, our study investigates how GenAI influences designers’ creativity in the context of pro-social advertising for products from low-income regions with minority populations. Specifically, we conceptualize creativity as a process encompassing two distinct stages: an ideation stage and an implementation stage. Our Study 1 (a lab experiment) demonstrates that GenAI tremendously enhances work creativity in the ideation stage by mitigating cognitive fixation for all designers. However, during the implementation stage, the impacts diverge: low-expertise designers with GenAI continue to experience improvements in work creativity, while high-expertise designers with GenAI show no gains in their work creativity and suffer a significant reduction in their work efficiency. Further video analyses reveal that expertise fixation underpins these impacts. That is, as GenAI introduces work approaches that deviate from high-expertise designers’ established approaches in implementation, co-creation with GenAI leads to counterproductive outcomes. Building on these findings, Study 2 (a field experiment) further validates the impact of GenAI and the role of expertise fixation among professional designers, and rules out alternative explanations. This study also employs a cutting-edge GenAI model to ensure the robustness of our findings against technological advancements. Our research advances the understanding of human-AI collaboration by highlighting GenAI’s nuanced role in the creative process and its varying effects based on designers’ expertise levels. Moreover, it offers actional insights for organizations, policymakers, and designers seeking to leverage AI for sustainable social impact. By examining creativity in the context of pro-social advertising, this study underscores the potential role of AI-driven innovation in design practices that benefit marginalized communities and support sustainable development.

Generative Artificial Intelligence (GenAI) is reshaping creative domains by offering novel and complex content solutions with transformative potential. However, the existing GenAI literature largely examines creativity as an end product, overlooking the intricate dynamics of the human-GenAI co-creative process. Addressing this gap, our study investigates how GenAI influences designers’ creativity in the context of pro-social advertising for products from low-income regions with minority populations. Specifically, we conceptualize creativity as a process encompassing two distinct stages: an ideation stage and an implementation stage. Our Study 1 (a lab experiment) demonstrates that GenAI tremendously enhances work creativity in the ideation stage by mitigating cognitive fixation for all designers. However, during the implementation stage, the impacts diverge: low-expertise designers with GenAI continue to experience improvements in work creativity, while high-expertise designers with GenAI show no gains in their work creativity and suffer a significant reduction in their work efficiency. Further video analyses reveal that expertise fixation underpins these impacts. That is, as GenAI introduces work approaches that deviate from high-expertise designers’ established approaches in implementation, co-creation with GenAI leads to counterproductive outcomes. Building on these findings, Study 2 (a field experiment) further validates the impact of GenAI and the role of expertise fixation among professional designers, and rules out alternative explanations. This study also employs a cutting-edge GenAI model to ensure the robustness of our findings against technological advancements. Our research advances the understanding of human-AI collaboration by highlighting GenAI’s nuanced role in the creative process and its varying effects based on designers’ expertise levels. Moreover, it offers actional insights for organizations, policymakers, and designers seeking to leverage AI for sustainable social impact. By examining creativity in the context of pro-social advertising, this study underscores the potential role of AI-driven innovation in design practices that benefit marginalized communities and support sustainable development.

Yumei He, Tulane University

Users of online dating platforms have been increasingly experiencing burnout—a state of psychological exhaustion characterized by emotional depletion, choice fatigue, and decreased engagement; this is partially due to the limitations of recommender systems that prioritize similarity-based matching. To mitigate user burnout, we propose a serendipity-oriented recommender system to strike a balance between unexpectedness and relevance in online dating recommendations, which we implement through deep learning attention networks that extract spiritual compatibility from user-generated texts while relaxing constraints on horizontal and vertical dating attributes. We evaluate this system through a randomized field experiment conducted by an online dating platform (N=11,919). Users in treatment group received serendipity-oriented recommendations, whereas those in control group were provided expertise-based recommendations. Our results reveal a trade-off between unexpectedness and engagement: users in the treatment group were 3.4% less likely to view profiles and 4.9% less likely to send requests, yet they achieved a 0.5% higher likelihood of forming matches. Further analyses suggest that the treatment effects are more pronounced for female users. More importantly, our findings provide suggestive evidence that serendipity mitigates user burnout, as users in treatment group significantly exhibit (1) a higher level of positive emotion in matching requests, indicating reduced emotional exhaustion, (2) relaxed sorting patterns on horizontal attributes (age, location, hometown), implying decreased choice fatigue, and (3) a greater likelihood of effective match and faster matching, reflecting enhanced engagement. Our study contributes to the literature on serendipity-based recommender system, dating preferences, market design of dating platforms. It also offers a scalable strategy for alleviating user burnout in dating platforms.

Users of online dating platforms have been increasingly experiencing burnout—a state of psychological exhaustion characterized by emotional depletion, choice fatigue, and decreased engagement; this is partially due to the limitations of recommender systems that prioritize similarity-based matching. To mitigate user burnout, we propose a serendipity-oriented recommender system to strike a balance between unexpectedness and relevance in online dating recommendations, which we implement through deep learning attention networks that extract spiritual compatibility from user-generated texts while relaxing constraints on horizontal and vertical dating attributes. We evaluate this system through a randomized field experiment conducted by an online dating platform (N=11,919). Users in treatment group received serendipity-oriented recommendations, whereas those in control group were provided expertise-based recommendations. Our results reveal a trade-off between unexpectedness and engagement: users in the treatment group were 3.4% less likely to view profiles and 4.9% less likely to send requests, yet they achieved a 0.5% higher likelihood of forming matches. Further analyses suggest that the treatment effects are more pronounced for female users. More importantly, our findings provide suggestive evidence that serendipity mitigates user burnout, as users in treatment group significantly exhibit (1) a higher level of positive emotion in matching requests, indicating reduced emotional exhaustion, (2) relaxed sorting patterns on horizontal attributes (age, location, hometown), implying decreased choice fatigue, and (3) a greater likelihood of effective match and faster matching, reflecting enhanced engagement. Our study contributes to the literature on serendipity-based recommender system, dating preferences, market design of dating platforms. It also offers a scalable strategy for alleviating user burnout in dating platforms.

PhD student in Operations Management, University of Texas at Dallas

Qiuxia (Katalia) Chen is a 4th year PhD student in Operations Management at the University of Texas at Dallas. Before starting her PhD, she earned a master’s degree in Supply Chain Management at the University of Texas at Dallas and a bachelor’s degree in Sociology (with the direction of Psychology) at Harbin Engineering University in China.

Katalia also had 4 years of industry experience before she pursued her master’s degree. She was in charge of the global business for a brand in AUKEY, a top 5 cross-boarder e-commerce company in China.

Katalia’s research interest lies in the sustainability, social impact in non-profit organizations, and behavioral operations management. Katalia is also a dog lover. She is currently working on two research projects about how to save dogs’ lives at animal shelters with Prof. Elena Katok and Prof. Ernan Haruvy. One project is utilizing the data provided by Dallas Animal Services, the only municipal shelter in city of Dallas.

Qiuxia (Katalia) Chen is a 4th year PhD student in Operations Management at the University of Texas at Dallas. Before starting her PhD, she earned a master’s degree in Supply Chain Management at the University of Texas at Dallas and a bachelor’s degree in Sociology (with the direction of Psychology) at Harbin Engineering University in China.

Katalia also had 4 years of industry experience before she pursued her master’s degree. She was in charge of the global business for a brand in AUKEY, a top 5 cross-boarder e-commerce company in China.

Katalia’s research interest lies in the sustainability, social impact in non-profit organizations, and behavioral operations management. Katalia is also a dog lover. She is currently working on two research projects about how to save dogs’ lives at animal shelters with Prof. Elena Katok and Prof. Ernan Haruvy. One project is utilizing the data provided by Dallas Animal Services, the only municipal shelter in city of Dallas.

Associate professor, Daniels School of Business, Purdue University

Gökçe Esenduran is an associate professor at Daniels School of Business, Purdue University. She received her Ph.D. from UNC at Chapel Hill. Before joining Purdue, she was an associate professor at The Ohio State University. Gökçe’s current research primarily focuses on environmentally and socially responsible operations, product returns, circular economy, and environmental regulations. She has published in journals such as Management Science, Manufacturing & Service Operations Management, Production and Operations Management, and Journal of Operations Management. She is a senior editor for Production and Operations Management and associate editor for Manufacturing & Service Operations Management and Decision Sciences Journal. In the past, she served as the treasurer of Women in OR/MS, as the secretary, president, and past-president of POMS College of Sustainable Operations, and as the chair of MSOM Sustainable Operations SIG. Currently, she is serving as the secretary/treasurer of the M&SOM Society Board and on INFORMS Magazine Editorial Advisory Board.

Gökçe Esenduran is an associate professor at Daniels School of Business, Purdue University. She received her Ph.D. from UNC at Chapel Hill. Before joining Purdue, she was an associate professor at The Ohio State University. Gökçe’s current research primarily focuses on environmentally and socially responsible operations, product returns, circular economy, and environmental regulations. She has published in journals such as Management Science, Manufacturing & Service Operations Management, Production and Operations Management, and Journal of Operations Management. She is a senior editor for Production and Operations Management and associate editor for Manufacturing & Service Operations Management and Decision Sciences Journal. In the past, she served as the treasurer of Women in OR/MS, as the secretary, president, and past-president of POMS College of Sustainable Operations, and as the chair of MSOM Sustainable Operations SIG. Currently, she is serving as the secretary/treasurer of the M&SOM Society Board and on INFORMS Magazine Editorial Advisory Board.

Assistant Professor, A.B. Freeman School of Business, Tulane University

Yumei He is an Assistant Professor at the A.B. Freeman School of Business, Tulane University. Her research focuses on human-AI interaction, generative AI adoption, and open-source AI models, with applications in online platforms such as dating and live streaming. She employs a range of methodologies, including randomized field experiments, econometrics, and machine learning. Dr. He’s work has been published in journals such as Information Systems Research and Journal of the Association for Information Systems.

In addition to her academic pursuits, she serves as an external researcher for various digital platforms and startups, bridging the gap between academia and industry. She has also received several awards, including the INFORMS eBusiness cluster best paper runner-up award, and actively contributes to the academic community as an Editorial Review Board member for Information Systems Research.

Yumei He is an Assistant Professor at the A.B. Freeman School of Business, Tulane University. Her research focuses on human-AI interaction, generative AI adoption, and open-source AI models, with applications in online platforms such as dating and live streaming. She employs a range of methodologies, including randomized field experiments, econometrics, and machine learning. Dr. He’s work has been published in journals such as Information Systems Researchand Journal of the Association for Information Systems.

In addition to her academic pursuits, she serves as an external researcher for various digital platforms and startups, bridging the gap between academia and industry. She has also received several awards, including the INFORMS eBusiness cluster best paper runner-up award, and actively contributes to the academic community as an Editorial Review Board member for Information Systems Research.

Associate Professor, A.B. Freeman School of Business, Tulane University

Yi-Jen (Ian) Ho is an associate professor at the A.B. Freeman School of Business, Tulane University. He is interested in understanding the impacts of emerging information technologies. His current research focuses on location-based services, artificial intelligence, and online platforms. He applies various methods to obtain insights and identify causalities, including game-theoretic modeling, econometrics, randomized experiments, and machine learning. His research has appeared in premier business journals, including Information Systems Research and Production and Operations Management. He received the Gordon B. Davis Young Scholar Award in 2022 and the Nunamaker-Chen Dissertation Award in 2017 from the INFORMS Information Systems Society. His research won several best paper awards and runners-ups at major conferences, including INFORMS Information Systems and eBusiness Sections, PACIS, and WeB. He has served as an associate editor for Management Information Systems Quarterly and a senior editor of special issues at Production and Operations Management. He also served as a cluster co-chair, associate editor, and program committee member for conferences.

Yi-Jen (Ian) Ho is an associate professor at the A.B. Freeman School of Business, Tulane University. He is interested in understanding the impacts of emerging information technologies. His current research focuses on location-based services, artificial intelligence, and online platforms. He applies various methods to obtain insights and identify causalities, including game-theoretic modeling, econometrics, randomized experiments, and machine learning. His research has appeared in premier business journals, including Information Systems Research and Production and Operations Management. He received the Gordon B. Davis Young Scholar Award in 2022 and the Nunamaker-Chen Dissertation Award in 2017 from the INFORMS Information Systems Society. His research won several best paper awards and runners-ups at major conferences, including INFORMS Information Systems and eBusiness Sections, PACIS, and WeB. He has served as an associate editor for Management Information Systems Quarterlyand a senior editor of special issues at Production and Operations Management. He also served as a cluster co-chair, associate editor, and program committee member for conferences.

Professor of Operations Management, Jindal School of Management, University of Texas at Dallas

Elena Katok joined the Jindal School of Management at the University of Texas at Dallas in 2012. She is Ashok and Monica Mago Professor of Operations Management. She is also on the International Faculty at the University of Cologne, Germany. Before her appointment at the University of Texas at Dallas, she was a Professor at the Smeal College of Business at Penn State, where she was a Zimmerman Faculty Fellow. She holds a Bachelor’s from the University of California, Berkeley, and an MBA and a Ph.D. degree from Penn State. Dr. Katok’s research is in the area of Behavioral Operations Management. She analyzes behavioral factors that affect supply chain contracts’ efficiency, procurement mechanism performance, and other channel coordination issues. Her work is published in Management Science, M&SOM, Production and Operations Management Journal, Journal of Operations Management, and other journals in business and economics. Dr. Katok was part of a team that won the 2000 Franz Edelman Award, the most prestigious award for the Practice of Operations Research and the Management Sciences. She is the Department Editor for the Operations Management Department at Management Science and an Associate Editor at M&SOM and the Production and Operations Management Journal. She served as a Department Editor at the Production and Operations Management Journal Behavioral Operation Department from 2012 to 2021. She also co-edited the Handbook of Behavioral Operations published by Wiley in 2018.

Elena Katok joined the Jindal School of Management at the University of Texas at Dallas in 2012. She is Ashok and Monica Mago Professor of Operations Management. She is also on the International Faculty at the University of Cologne, Germany. Before her appointment at the University of Texas at Dallas, she was a Professor at the Smeal College of Business at Penn State, where she was a Zimmerman Faculty Fellow. She holds a Bachelor’s from the University of California, Berkeley, and an MBA and a Ph.D. degree from Penn State. Dr. Katok’s research is in the area of Behavioral Operations Management. She analyzes behavioral factors that affect supply chain contracts’ efficiency, procurement mechanism performance, and other channel coordination issues. Her work is published in Management Science, M&SOM, Production and Operations Management Journal, Journal of Operations Management, and other journals in business and economics. Dr. Katok was part of a team that won the 2000 Franz Edelman Award, the most prestigious award for the Practice of Operations Research and the Management Sciences. She is the Department Editor for the Operations Management Department at Management Science and an Associate Editor at M&SOM and the Production and Operations Management Journal. She served as a Department Editor at the Production and Operations Management Journal Behavioral Operation Department from 2012 to 2021. She also co-edited the Handbook of Behavioral Operations published by Wiley in 2018.

Professor of information systems and chair of the information systems and operations management area, Costello College of Business, George Mason University

Pallab Sanyal is a professor of information systems and the chair of the information systems and operations management (ISOM) Area at the Costello College of Business, George Mason University. Sanyal’s primary research interest lies in understanding how the design of information systems influences the decisions people make, leading to certain individual, organizational, and societal outcomes. His research has been published in Management Science, MIS Quarterly, Information Systems Research (ISR), the Journal of Management Information Systems (JMIS), Production and Operations Management (POM), and the Journal of Operations Management (JOM) among other outlets. He has presented his research at many international conferences, including CIST, ICIS, INFORMS, SCECR, WISE, and WITS. At WITS, his papers were twice nominated for the best paper award. Sanyal’s dissertation research won the Design Science Research Award from the INFORMS Information Systems Society. Sanyal has been serving as an Associate Editor (AE) of ISR since 2018. Additionally, he serves as an Editorial Review Board Member at the Journal of the Association for Information Systems (JAIS), where he previously held the role of AE for three years. He has received the Best Reviewer and Best AE awards from ISR. He has also served on the organization committees of many conferences, including AMCIS and WITS.

Pallab Sanyal is a professor of information systems and the chair of the information systems and operations management (ISOM) Area at the Costello College of Business, George Mason University. Sanyal’s primary research interest lies in understanding how the design of information systems influences the decisions people make, leading to certain individual, organizational, and societal outcomes. His research has been published in Management Science, MIS Quarterly, Information Systems Research (ISR), the Journal of Management Information Systems (JMIS), Production and Operations Management (POM), and the Journal of Operations Management (JOM) among other outlets. He has presented his research at many international conferences, including CIST, ICIS, INFORMS, SCECR, WISE, and WITS. At WITS, his papers were twice nominated for the best paper award. Sanyal’s dissertation research won the Design Science Research Award from the INFORMS Information Systems Society. Sanyal has been serving as an Associate Editor (AE) of ISR since 2018. Additionally, he serves as an Editorial Review Board Member at the Journal of the Association for Information Systems (JAIS), where he previously held the role of AE for three years. He has received the Best Reviewer and Best AE awards from ISR. He has also served on the organization committees of many conferences, including AMCIS and WITS.

Assistant Professor of Operations, Information and Decisions, The Wharton School, University of Pennsylvania

Neha Sharma is an Assistant Professor of Operations, Information and Decisions at The Wharton School, University of Pennsylvania. Her research uses data driven analytical models to study pricing and incentives in marketplaces. Her work focusses on designing two sided platforms where users share assets like cars, or property, and platforms with no tangible assets such as knowledge sharing communities. She has been a finalist in many paper competitions like Service Science cluster and Service Science student competitions.

Neha Sharma is an Assistant Professor of Operations, Information and Decisions at The Wharton School, University of Pennsylvania. Her research uses data driven analytical models to study pricing and incentives in marketplaces. Her work focusses on designing two sided platforms where users share assets like cars, or property, and platforms with no tangible assets such as knowledge sharing communities. She has been a finalist in many paper competitions like Service Science cluster and Service Science student competitions.

Associate Professor of Operations, McCombs School of Business, The University of Texas at Austin

Ioannis (Yannis) Stamatopoulus is an Associate Professor of Operations at The University of Texas at Austin’s McCombs School of Business. He teaches experiential operations classes at the MBA level and studies technology and operations, revenue management, and public policy. His works have appeared in Science, Operations Research, Management Science, Marketing Science, Manufacturing and Service Operations Management, and Production and Operations Management, and have been discussed, among other places, by the NY Times, The Associated Press, The Washington Post, Bloomberg, and the NPR.

Ioannis (Yannis) Stamatopoulus is an Associate Professor of Operations at The University of Texas at Austin’s McCombs School of Business. He teaches experiential operations classes at the MBA level and studies technology and operations, revenue management, and public policy. His works have appeared in Science, Operations Research, Management Science, Marketing Science, Manufacturing and Service Operations Management, and Production and Operations Management, and have been discussed, among other places, by the NY Times, The Associated Press, The Washington Post, Bloomberg, and the NPR.

Professor of Operations Management, Scheller College of Business, Georgia Tech

Ravi Subramanian is Gregory J. Owens Professor of Operations Management at the Scheller College of Business at Georgia Tech. Ravi’s research focuses on environmental and social sustainability and interfaces operations and supply chain management with a range of disciplines. Ravi’s published research spans three categories: (i) environmental sustainability in firms and supply chains; (ii) value and effectiveness of corporate sustainability efforts; and, (iii) transparency and disclosure. More recently, he has extended his research along themes centered on socioeconomic impact. Ravi received the inaugural Paul Kleindorfer Award in Sustainability in 2012.

Ravi’s teaching interests are in data analytics, operations/supply chain management, and business sustainability, with a focus on the interrelationships among economic, environmental, and social dimensions of organizational performance. Ravi received the Scheller College of Business’ Teaching Excellence Awards in 2009, 2014, 2020, and 2024. Ravi served as Faculty Director of Georgia Tech’s flagship Denning Technology and Management Program from 2013 to 2016.

Ravi has prior business experience in operations strategy, manufacturing planning, and ERP implementation. He has also collaborated on NSF SBIR-funded research with UrjaNet (acquired by Arcadia in 2022). He holds a PhD in Operations and Management Science (Ross) and an MS in Industrial and Operations Engineering from the University of Michigan Ann Arbor, a Master of Management degree from IIT Mumbai (India), and a Bachelor’s degree in Mechanical Engineering from BITS Pilani (India).

Ravi Subramanian is Gregory J. Owens Professor of Operations Management at the Scheller College of Business at Georgia Tech. Ravi’s research focuses on environmental and social sustainability and interfaces operations and supply chain management with a range of disciplines. Ravi’s published research spans three categories: (i) environmental sustainability in firms and supply chains; (ii) value and effectiveness of corporate sustainability efforts; and, (iii) transparency and disclosure. More recently, he has extended his research along themes centered on socioeconomic impact. Ravi received the inaugural Paul Kleindorfer Award in Sustainability in 2012.

Ravi’s teaching interests are in data analytics, operations/supply chain management, and business sustainability, with a focus on the interrelationships among economic, environmental, and social dimensions of organizational performance. Ravi received the Scheller College of Business’ Teaching Excellence Awards in 2009, 2014, 2020, and 2024. Ravi served as Faculty Director of Georgia Tech’s flagship Denning Technology and Management Program from 2013 to 2016.

Ravi has prior business experience in operations strategy, manufacturing planning, and ERP implementation. He has also collaborated on NSF SBIR-funded research with UrjaNet (acquired by Arcadia in 2022). He holds a PhD in Operations and Management Science (Ross) and an MS in Industrial and Operations Engineering from the University of Michigan Ann Arbor, a Master of Management degree from IIT Mumbai (India), and a Bachelor’s degree in Mechanical Engineering from BITS Pilani (India).

Professor of Operations Management, Scheller College of Business, Georgia Tech

Gang Wang is an associate professor of Management Information Systems (MIS) and Joint Educational Institute (JEI) research fellow at the Lerner College of Business & Economics, the University of Delaware. He also serves as a co-director of the FinTech Innovation Hub, a university-level initiative. He received his Ph.D. degree in Operations and Information Management from the University of Connecticut. His research interests include e-platforms, digital technology adoption, and social impact of technology. His research has been published in Management Science, MIS Quarterly, Information Systems Research and other premier academic journals. He currently serves as an associate editor at Decision Support Systems, and Decision Science. He also served as a guest senior editor at Production and Operations Management.

Gang Wang is an associate professor of Management Information Systems (MIS) and Joint Educational Institute (JEI) research fellow at the Lerner College of Business & Economics, the University of Delaware. He also serves as a co-director of the FinTech Innovation Hub, a university-level initiative. He received his Ph.D. degree in Operations and Information Management from the University of Connecticut. His research interests include e-platforms, digital technology adoption, and social impact of technology. His research has been published in Management Science, MIS Quarterly, Information Systems Research and other premier academic journals. He currently serves as an associate editor at Decision Support Systems, and Decision Science. He also served as a guest senior editor at Production and Operations Management.

Professor and associate professor of Management Information Systems, Terry College of Business, University of Georgia

Ling Xue is the Terry Alumni Board Distinguished Professor and an associate professor of Management Information Systems at Terry College of Business, University of Georgia. He earned his PhD in Information Systems from the University of Texas at Austin.

His research expertise lies at the intersection of digital technology and decentralized governance. He delves into the socio-technical aspects within various digital contexts, such as digital platforms, open-source communities, blockchain ecosystems, AI development and deployment phenomena, and traditional corporate IT environments. His research has been published in research journals in various business disciplines, including Information Systems Research (ISR), Management Information Systems Quarterly (MISQ), Academy of Management Journal (AMJ), Journal of Operations Management (JOM), Production and Operations Management (POM), Journal of Management Information Systems (MIS), Journal of Information Systems (JIS), Decision Support Systems (DSS), Information & Management (I&M), International Journal of Electronic Commerce (IJEC) etc.

He is currently an associate editor at ISR, a senior editor at POM, an associate editor at Journal of Electronic Commerce Research (JECR), and on the Editorial Review Board of Management Science (MS). He was an associate editor of MISQ. He has also served as the president and the secretary of treasury of E-Business Section of INFORMS. He was a recipient of multiple awards, including the Best Senior Editor of POM Society, the AIS (Association of Information Systems) Mid-Career Award, the reviewer of the year from ISR, and the faculty research recognition award from J Mack Robinson College of Business at Georgia State University. Before joining the University of Georgia, he had served as a faculty member at Georgia State University, University of North Carolina at Greensboro, University of Memphis, and University of Scranton.

Ling Xue is the Terry Alumni Board Distinguished Professor and an associate professor of Management Information Systems at Terry College of Business, University of Georgia. He earned his PhD in Information Systems from the University of Texas at Austin.

His research expertise lies at the intersection of digital technology and decentralized governance. He delves into the socio-technical aspects within various digital contexts, such as digital platforms, open-source communities, blockchain ecosystems, AI development and deployment phenomena, and traditional corporate IT environments. His research has been published in research journals in various business disciplines, including Information Systems Research (ISR), Management Information Systems Quarterly (MISQ), Academy of Management Journal (AMJ), Journal of Operations Management (JOM), Production and Operations Management (POM), Journal of Management Information Systems (MIS), Journal of Information Systems (JIS), Decision Support Systems (DSS), Information & Management (I&M), International Journal of Electronic Commerce (IJEC) etc.

He is currently an associate editor at ISR, a senior editor at POM, an associate editor at Journal of Electronic Commerce Research (JECR), and on the Editorial Review Board of Management Science (MS). He was an associate editor of MISQ. He has also served as the president and the secretary of treasury of E-Business Section of INFORMS. He was a recipient of multiple awards, including the Best Senior Editor of POM Society, the AIS (Association of Information Systems) Mid-Career Award, the reviewer of the year from ISR, and the faculty research recognition award from J Mack Robinson College of Business at Georgia State University. Before joining the University of Georgia, he had served as a faculty member at Georgia State University, University of North Carolina at Greensboro, University of Memphis, and University of Scranton.