Mary Czerwinski
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Mary Czerwinski has a total of 137 co-authors. Here are the top co-authors by number of shared publications:
- Jina Suh
University of Washington, Seattle, USA
11 shared publications - Javier Hernandez
Microsoft Research, Cambridge, Massachusetts, USA
9 shared publications - Koustuv Saha
Siebel School of Computing and Data Science, University of Illinois Urbana-Champaign, Urbana, Illinois, USA
7 shared publications - Daniel McDuff
Google Research, Mountain View, USA
6 shared publications - Shamsi Iqbal
Microsoft Research, Redmond, WA, USA
6 shared publications - Subigya Nepal
Department of Computer Science, University of Virginia
5 shared publications
Latest Publications
- Open Access
- 16 October 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (7) , 1-41
From User Surveys to Telemetry-Driven AI Agents: Exploring the Potential of Personalized Productivity Solutions
Information workers increasingly struggle with productivity challenges in modern workplaces, facing difficulties in managing time and effectively utilizing workplace analytics data for behavioral improvement. Despite the availability of productivity metrics through enterprise tools, workers often fail to translate this data into actionable insights. We present a comprehensive, user-centric approach to address these challenges through AI-based productivity agents tailored to users' needs. Utilizing a two-phase method, we first conducted a survey with 363 participants, exploring various aspects of productivity, communication style, agent approach, personality traits, personalization, and privacy. Drawing on the survey insights, we developed a GPT-4 powered personalized productivity agent that utilizes telemetry data gathered via Viva Insights from information workers to provide tailored assistance. We compared its performance with alternative productivity-assistive tools, such as dashboard and narrative, in a study involving 40 participants. Our findings highlight the importance of user-centric design, adaptability, and the balance between personalization and privacy in AI-assisted productivity tools. By building on these insights, our work provides important guidance for developing more effective productivity solutions, ultimately leading to optimized efficiency and user experiences for information workers.
- 16 October 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (7) , 1-41
Open AccessFrom User Surveys to Telemetry-Driven AI Agents: Exploring the Potential of Personalized Productivity Solutions
Information workers increasingly struggle with productivity challenges in modern workplaces, facing difficulties in managing time and effectively utilizing workplace analytics data for behavioral improvement. Despite the availability of productivity metrics through enterprise tools, workers often fail to translate this data into actionable insights. We present a comprehensive, user-centric approach to address these challenges through AI-based productivity agents tailored to users' needs. Utilizing a two-phase method, we first conducted a survey with 363 participants, exploring various aspects of productivity, communication style, agent approach, personality traits, personalization, and privacy. Drawing on the survey insights, we developed a GPT-4 powered personalized productivity agent that utilizes telemetry data gathered via Viva Insights from information workers to provide tailored assistance. We compared its performance with alternative productivity-assistive tools, such as dashboard and narrative, in a study involving 40 participants. Our findings highlight the importance of user-centric design, adaptability, and the balance between personalization and privacy in AI-assisted productivity tools. By building on these insights, our work provides important guidance for developing more effective productivity solutions, ultimately leading to optimized efficiency and user experiences for information workers.
- Open Access
Affective Air Quality Dataset: Personal Chemical Emissions from Emotional Videos
Inspired by the role of chemosignals in conveying emotional states, this paper introduces the Affective Air Quality (AAQ) dataset, a novel dataset collected to explore the potential of volatile odor compound and gas sensor data for non-contact emotion detection. This dataset bridges the gap between the realms of breath \& body odor emission (personal chemical emissions) analysis and established practices in affective computing. Comprising 4-channel gas sensor data from 23 participants at two distances from the body (wearable and desktop), alongside emotional ratings elicited by targeted movie clips, the dataset encapsulates initial groundwork to analyze the correlation between personal chemical emissions and varied emotional responses. The AAQ dataset also provides insights drawn from exit interviews, thereby painting a holistic picture of perceptions regarding air quality monitoring and its implications for privacy. By offering this dataset alongside preliminary attempts at emotion recognition models based on it to the broader research community, we seek to advance the development of odor-based affect recognition models that prioritize user privacy and comfort.
Open AccessAffective Air Quality Dataset: Personal Chemical Emissions from Emotional Videos
Inspired by the role of chemosignals in conveying emotional states, this paper introduces the Affective Air Quality (AAQ) dataset, a novel dataset collected to explore the potential of volatile odor compound and gas sensor data for non-contact emotion detection. This dataset bridges the gap between the realms of breath \& body odor emission (personal chemical emissions) analysis and established practices in affective computing. Comprising 4-channel gas sensor data from 23 participants at two distances from the body (wearable and desktop), alongside emotional ratings elicited by targeted movie clips, the dataset encapsulates initial groundwork to analyze the correlation between personal chemical emissions and varied emotional responses. The AAQ dataset also provides insights drawn from exit interviews, thereby painting a holistic picture of perceptions regarding air quality monitoring and its implications for privacy. By offering this dataset alongside preliminary attempts at emotion recognition models based on it to the broader research community, we seek to advance the development of odor-based affect recognition models that prioritize user privacy and comfort.
- Open Access
- 2 May 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (2) , 1-45
AURA: Amplifying Understanding, Resilience, and Awareness for Responsible AI Content Work
Behind the scenes of maintaining the safety of technology products from harmful and illegal digital content lies unrecognized human labor. The recent rise in the use of generative AI technologies and the accelerating demands to meet responsible AI (RAI) aims necessitates an increased focus on the labor behind such efforts in the age of AI. This study investigates the nature and challenges of content work that supports RAI efforts, or "RAI content work," that spans content moderation, data labeling, and red teaming -- through the lived experiences of content workers. We conduct a formative survey and semi-structured interview studies to develop a conceptualization of RAI content work and a subsequent framework of recommendations for providing holistic support for content workers. We validate our recommendations through a series of workshops with content workers and derive considerations for and examples of implementing such recommendations. We discuss how our framework may guide future innovation to support the well-being and professional development of the RAI content workforce.
- 2 May 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (2) , 1-45
Open AccessAURA: Amplifying Understanding, Resilience, and Awareness for Responsible AI Content Work
Behind the scenes of maintaining the safety of technology products from harmful and illegal digital content lies unrecognized human labor. The recent rise in the use of generative AI technologies and the accelerating demands to meet responsible AI (RAI) aims necessitates an increased focus on the labor behind such efforts in the age of AI. This study investigates the nature and challenges of content work that supports RAI efforts, or "RAI content work," that spans content moderation, data labeling, and red teaming -- through the lived experiences of content workers. We conduct a formative survey and semi-structured interview studies to develop a conceptualization of RAI content work and a subsequent framework of recommendations for providing holistic support for content workers. We validate our recommendations through a series of workshops with content workers and derive considerations for and examples of implementing such recommendations. We discuss how our framework may guide future innovation to support the well-being and professional development of the RAI content workforce.
- Top Cited PapersOpen Access
- 25 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-29
AI on My Shoulder: Supporting Emotional Labor in Front-Office Roles with an LLM-based Empathetic Coworker
- 25 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-29
Top Cited PapersOpen AccessAI on My Shoulder: Supporting Emotional Labor in Front-Office Roles with an LLM-based Empathetic Coworker
- 22 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51 (2) , 344-354
Triple Peak Day: Work Rhythms of Software Developers in Hybrid Work
The future of work is rapidly changing, with remote and hybrid settings blurring the boundaries between professional and personal life. To understand how work rhythms vary across different work settings, we conducted a month-long study of 65 software developers, collecting anonymized computer activity data as well as daily ratings for perceived stress, productivity, and work setting. In addition to confirming the double-peak pattern of activity at 10:00 am and 2:00 pm observed in prior research, we observed a significant third peak around 9:00 pm. This third peak was associated with higher perceived productivity during remote days but increased stress during onsite and hybrid days, highlighting a nuanced interplay between work demands and work settings. Additionally, we found strong correlations between computer activity, productivity, and stress, including an inverted U-shaped relationship where productivity peaked at around six hours of computer activity before declining on more active days. These findings provide new insights into evolving work rhythms and highlight the impact of different work settings on productivity and stress.
- 22 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 51 (2) , 344-354
Triple Peak Day: Work Rhythms of Software Developers in Hybrid Work
The future of work is rapidly changing, with remote and hybrid settings blurring the boundaries between professional and personal life. To understand how work rhythms vary across different work settings, we conducted a month-long study of 65 software developers, collecting anonymized computer activity data as well as daily ratings for perceived stress, productivity, and work setting. In addition to confirming the double-peak pattern of activity at 10:00 am and 2:00 pm observed in prior research, we observed a significant third peak around 9:00 pm. This third peak was associated with higher perceived productivity during remote days but increased stress during onsite and hybrid days, highlighting a nuanced interplay between work demands and work settings. Additionally, we found strong correlations between computer activity, productivity, and stress, including an inverted U-shaped relationship where productivity peaked at around six hours of computer activity before declining on more active days. These findings provide new insights into evolving work rhythms and highlight the impact of different work settings on productivity and stress.
- Top Cited PapersOpen Access
- 21 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 8 (4) , 1-44
MindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences
Mental health concerns are prevalent among college students, highlighting the need for effective interventions that promote self-awareness and holistic well-being. MindScape explores a novel approach to AI-powered journaling by integrating passively collected behavioral patterns such as conversational engagement, sleep, and location with Large Language Models (LLMs). This integration creates a highly personalized and context-aware journaling experience, enhancing self-awareness and well-being by embedding behavioral intelligence into AI. We present an 8-week exploratory study with 20 college students, demonstrating the MindScape app's efficacy in enhancing positive affect (7%), reducing negative affect (11%), loneliness (6%), and anxiety and depression, with a significant week-over-week decrease in PHQ-4 scores (-0.25 coefficient). The study highlights the advantages of contextual AI journaling, with participants particularly appreciating the tailored prompts and insights provided by the MindScape app. Our analysis also includes a comparison of responses to AI-driven contextual versus generic prompts, participant feedback insights, and proposed strategies for leveraging contextual AI journaling to improve well-being on college campuses. By showcasing the potential of contextual AI journaling to support mental health, we provide a foundation for further investigation into the effects of contextual AI journaling on mental health and well-being.
- 21 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 8 (4) , 1-44
Top Cited PapersOpen AccessMindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences
Mental health concerns are prevalent among college students, highlighting the need for effective interventions that promote self-awareness and holistic well-being. MindScape explores a novel approach to AI-powered journaling by integrating passively collected behavioral patterns such as conversational engagement, sleep, and location with Large Language Models (LLMs). This integration creates a highly personalized and context-aware journaling experience, enhancing self-awareness and well-being by embedding behavioral intelligence into AI. We present an 8-week exploratory study with 20 college students, demonstrating the MindScape app's efficacy in enhancing positive affect (7%), reducing negative affect (11%), loneliness (6%), and anxiety and depression, with a significant week-over-week decrease in PHQ-4 scores (-0.25 coefficient). The study highlights the advantages of contextual AI journaling, with participants particularly appreciating the tailored prompts and insights provided by the MindScape app. Our analysis also includes a comparison of responses to AI-driven contextual versus generic prompts, participant feedback insights, and proposed strategies for leveraging contextual AI journaling to improve well-being on college campuses. By showcasing the potential of contextual AI journaling to support mental health, we provide a foundation for further investigation into the effects of contextual AI journaling on mental health and well-being.
- 15 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 378-393
AImagery: A Multisensory Approach to Anxiety Reduction with AI, Olfactory Stimuli, and Biofeedback-Enhanced Guided Imagery
We present AImagery, an AI-powered immersive relaxation experience tailored to user preferences and physiological feedback. In a study with 32 participants, half of them experienced a multisensory experience with scent, biofeedback, and a personalized AI audio story based on their heart rate, self-reported mood and custom scenery. The experimental group showed a significant anxiety reduction for those with moderate to high anxiety, as opposed to the control group (non-guided meditation/baseline resting control condition). User feedback was positive and received high ratings for enjoyment, immersion and, to a lesser extend, sleepiness. Our results highlight the potential of multisensory AI-driven relaxation tools for those with elevated anxiety.
- 15 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 378-393
AImagery: A Multisensory Approach to Anxiety Reduction with AI, Olfactory Stimuli, and Biofeedback-Enhanced Guided Imagery
We present AImagery, an AI-powered immersive relaxation experience tailored to user preferences and physiological feedback. In a study with 32 participants, half of them experienced a multisensory experience with scent, biofeedback, and a personalized AI audio story based on their heart rate, self-reported mood and custom scenery. The experimental group showed a significant anxiety reduction for those with moderate to high anxiety, as opposed to the control group (non-guided meditation/baseline resting control condition). User feedback was positive and received high ratings for enjoyment, immersion and, to a lesser extend, sleepiness. Our results highlight the potential of multisensory AI-driven relaxation tools for those with elevated anxiety.
- Open Access
- 12 September 2024
- Published by JMIR Publications Inc. in JMIR Mental Health
- Vol. 11 (1) , e48974
Toward Tailoring Just-in-Time Adaptive Intervention Systems for Workplace Stress Reduction: Exploratory Analysis of Intervention Implementation
Background: Integrating stress-reduction interventions into the workplace may improve the health and well-being of employees, and there is an opportunity to leverage ubiquitous everyday work technologies to understand dynamic work contexts and facilitate stress reduction wherever work happens. Sensing-powered just-in-time adaptive intervention (JITAI) systems have the potential to adapt and deliver tailored interventions, but such adaptation requires a comprehensive analysis of contextual and individual-level variables that may influence intervention outcomes and be leveraged to drive the system’s decision-making. Objective: This study aims to identify key tailoring variables that influence momentary engagement in digital stress reduction microinterventions to inform the design of similar JITAI systems. Methods: To inform the design of such dynamic adaptation, we analyzed data from the implementation and deployment of a system that incorporates passively sensed data across everyday work devices to send just-in-time stress reduction microinterventions in the workplace to 43 participants during a 4-week deployment. We evaluated 27 trait-based factors (ie, individual characteristics), state-based factors (ie, workplace contextual and behavioral signals and momentary stress), and intervention-related factors (ie, location and function) across 1585 system-initiated interventions. We built logistical regression models to identify the factors contributing to momentary engagement, the choice of interventions, the engagement given an intervention choice, the user rating of interventions engaged, and the stress reduction from the engagement. Results: We found that women (odds ratio [OR] 0.41, 95% CI 0.21-0.77; P=.03), those with higher neuroticism (OR 0.57, 95% CI 0.39-0.81; P=.01), those with higher cognitive reappraisal skills (OR 0.69, 95% CI 0.52-0.91; P=.04), and those that chose calm interventions (OR 0.43, 95% CI 0.23-0.78; P=.03) were significantly less likely to experience stress reduction, while those with higher agreeableness (OR 1.73, 95% CI 1.10-2.76; P=.06) and those that chose prompt-based (OR 6.65, 95% CI 1.53-36.45; P=.06) or video-based (OR 5.62, 95% CI 1.12-34.10; P=.12) interventions were substantially more likely to experience stress reduction. We also found that work-related contextual signals such as higher meeting counts (OR 0.62, 95% CI 0.49-0.78; P<.001) and higher engagement skewness (OR 0.64, 95% CI 0.51-0.79; P<.001) were associated with a lower likelihood of engagement, indicating that state-based contextual factors such as being in a meeting or the time of the day may matter more for engagement than efficacy. In addition, a just-in-time intervention that was explicitly rescheduled to a later time was more likely to be engaged with (OR 1.77, 95% CI 1.32-2.38; P<.001). Conclusions: JITAI systems have the potential to integrate timely support into the workplace. On the basis of our findings, we recommend that individual, contextual, and content-based factors be incorporated into the system for tailoring as well as for monitoring ineffective engagements across subgroups and contexts.
- 12 September 2024
- Published by JMIR Publications Inc. in JMIR Mental Health
- Vol. 11 (1) , e48974
Open AccessToward Tailoring Just-in-Time Adaptive Intervention Systems for Workplace Stress Reduction: Exploratory Analysis of Intervention Implementation
Background: Integrating stress-reduction interventions into the workplace may improve the health and well-being of employees, and there is an opportunity to leverage ubiquitous everyday work technologies to understand dynamic work contexts and facilitate stress reduction wherever work happens. Sensing-powered just-in-time adaptive intervention (JITAI) systems have the potential to adapt and deliver tailored interventions, but such adaptation requires a comprehensive analysis of contextual and individual-level variables that may influence intervention outcomes and be leveraged to drive the system’s decision-making. Objective: This study aims to identify key tailoring variables that influence momentary engagement in digital stress reduction microinterventions to inform the design of similar JITAI systems. Methods: To inform the design of such dynamic adaptation, we analyzed data from the implementation and deployment of a system that incorporates passively sensed data across everyday work devices to send just-in-time stress reduction microinterventions in the workplace to 43 participants during a 4-week deployment. We evaluated 27 trait-based factors (ie, individual characteristics), state-based factors (ie, workplace contextual and behavioral signals and momentary stress), and intervention-related factors (ie, location and function) across 1585 system-initiated interventions. We built logistical regression models to identify the factors contributing to momentary engagement, the choice of interventions, the engagement given an intervention choice, the user rating of interventions engaged, and the stress reduction from the engagement. Results: We found that women (odds ratio [OR] 0.41, 95% CI 0.21-0.77; P=.03), those with higher neuroticism (OR 0.57, 95% CI 0.39-0.81; P=.01), those with higher cognitive reappraisal skills (OR 0.69, 95% CI 0.52-0.91; P=.04), and those that chose calm interventions (OR 0.43, 95% CI 0.23-0.78; P=.03) were significantly less likely to experience stress reduction, while those with higher agreeableness (OR 1.73, 95% CI 1.10-2.76; P=.06) and those that chose prompt-based (OR 6.65, 95% CI 1.53-36.45; P=.06) or video-based (OR 5.62, 95% CI 1.12-34.10; P=.12) interventions were substantially more likely to experience stress reduction. We also found that work-related contextual signals such as higher meeting counts (OR 0.62, 95% CI 0.49-0.78; P<.001) and higher engagement skewness (OR 0.64, 95% CI 0.51-0.79; P<.001) were associated with a lower likelihood of engagement, indicating that state-based contextual factors such as being in a meeting or the time of the day may matter more for engagement than efficacy. In addition, a just-in-time intervention that was explicitly rescheduled to a later time was more likely to be engaged with (OR 1.77, 95% CI 1.32-2.38; P<.001). Conclusions: JITAI systems have the potential to integrate timely support into the workplace. On the basis of our findings, we recommend that individual, contextual, and content-based factors be incorporated into the system for tailoring as well as for monitoring ineffective engagements across subgroups and contexts.
- Top Cited PapersOpen Access
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- Vol. 2024, 1-8
Contextual AI Journaling: Integrating LLM and Time Series Behavioral Sensing Technology to Promote Self-Reflection and Well-being using the MindScape App
MindScape aims to study the benefits of integrating time series behavioral patterns (e.g., conversational engagement, sleep, location) with Large Language Models (LLMs) to create a new form of contextual AI journaling, promoting self-reflection and well-being. We argue that integrating behavioral sensing in LLMs will likely lead to a new frontier in AI. In this Late-Breaking Work paper, we discuss the MindScape contextual journal App design that uses LLMs and behavioral sensing to generate contextual and personalized journaling prompts crafted to encourage self-reflection and emotional development. We also discuss the MindScape study of college students based on a preliminary user study and our upcoming study to assess the effectiveness of contextual AI journaling in promoting better well-being on college campuses. MindScape represents a new application class that embeds behavioral intelligence in AI.
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- Vol. 2024, 1-8
Top Cited PapersOpen AccessContextual AI Journaling: Integrating LLM and Time Series Behavioral Sensing Technology to Promote Self-Reflection and Well-being using the MindScape App
MindScape aims to study the benefits of integrating time series behavioral patterns (e.g., conversational engagement, sleep, location) with Large Language Models (LLMs) to create a new form of contextual AI journaling, promoting self-reflection and well-being. We argue that integrating behavioral sensing in LLMs will likely lead to a new frontier in AI. In this Late-Breaking Work paper, we discuss the MindScape contextual journal App design that uses LLMs and behavioral sensing to generate contextual and personalized journaling prompts crafted to encourage self-reflection and emotional development. We also discuss the MindScape study of college students based on a preliminary user study and our upcoming study to assess the effectiveness of contextual AI journaling in promoting better well-being on college campuses. MindScape represents a new application class that embeds behavioral intelligence in AI.
- Top Cited PapersOpen Access
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- Vol. 2024, 1-18
Understanding the Role of Large Language Models in Personalizing and Scaffolding Strategies to Combat Academic Procrastination
Traditional interventions for academic procrastination often fail to capture the nuanced, individual-specific factors that underlie them. Large language models (LLMs) hold immense potential for addressing this gap by permitting open-ended inputs, including the ability to customize interventions to individuals’ unique needs. However, user expectations and potential limitations of LLMs in this context remain underexplored. To address this, we conducted interviews and focus group discussions with 15 university students and 6 experts, during which a technology probe for generating personalized advice for managing procrastination was presented. Our results highlight the necessity for LLMs to provide structured, deadline-oriented steps and enhanced user support mechanisms. Additionally, our results surface the need for an adaptive approach to questioning based on factors like busyness. These findings offer crucial design implications for the development of LLM-based tools for managing procrastination while cautioning the use of LLMs for therapeutic guidance.
- 11 May 2024
- Published by Association for Computing Machinery (ACM)
- Vol. 2024, 1-18
Top Cited PapersOpen AccessUnderstanding the Role of Large Language Models in Personalizing and Scaffolding Strategies to Combat Academic Procrastination
Traditional interventions for academic procrastination often fail to capture the nuanced, individual-specific factors that underlie them. Large language models (LLMs) hold immense potential for addressing this gap by permitting open-ended inputs, including the ability to customize interventions to individuals’ unique needs. However, user expectations and potential limitations of LLMs in this context remain underexplored. To address this, we conducted interviews and focus group discussions with 15 university students and 6 experts, during which a technology probe for generating personalized advice for managing procrastination was presented. Our results highlight the necessity for LLMs to provide structured, deadline-oriented steps and enhanced user support mechanisms. Additionally, our results surface the need for an adaptive approach to questioning based on factors like busyness. These findings offer crucial design implications for the development of LLM-based tools for managing procrastination while cautioning the use of LLMs for therapeutic guidance.