Amal Hashky
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Amal Hashky has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Eric D. Ragan
University of Florida, USA
2 shared publications Eric D. Ragan
1 shared publication- Jaime Ruiz
University of Florida, Gainesville, Florida, USA
1 shared publication - Marcin Karcz
Department of Computer and Information Science and Engineering, University of Florida, Gainesville, Florida, USA
1 shared publication Neha Rani
1 shared publication- Reza Shahriari
University of Florida, USA
1 shared publication
Latest Publications
- Open Access
- 25 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 16 (3) , 1-23
Comparison of Text-Based Inputs for Human-in-the-Loop Feedback in Vision–Language Models
Human-in-the-loop methods leverage human feedback to enhance machine learning and artificial intelligence. Manual review of outputs can correct errors, identify model weaknesses, or expand labels to broaden model capabilities. Feedback collection methods range from simple flagging of outputs as correct or incorrect to more complex feature-level adjustments or natural language interpretations. This paper presents a user study evaluating changes in user performance over time and explores the trade-off between feedback quality and human effort. We compare four interactive input methods for reviewing and correcting outcomes in object detection and activity recognition in videos. Our findings indicate that while some complex input methods, such as free-text, require more time, the quality and impact of their feedback on model accuracy often surpass those of simpler methods that require less effort. However, more effort does not always lead to better-quality feedback, especially when aiming to improve the model. Our VLM experiments show that the most accurate models were trained using detailed natural language feedback or precise word-level corrections, while simple yes/no judgments also led to solid performance at a much lower annotation cost.
- 25 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 16 (3) , 1-23
Open AccessComparison of Text-Based Inputs for Human-in-the-Loop Feedback in Vision–Language Models
Human-in-the-loop methods leverage human feedback to enhance machine learning and artificial intelligence. Manual review of outputs can correct errors, identify model weaknesses, or expand labels to broaden model capabilities. Feedback collection methods range from simple flagging of outputs as correct or incorrect to more complex feature-level adjustments or natural language interpretations. This paper presents a user study evaluating changes in user performance over time and explores the trade-off between feedback quality and human effort. We compare four interactive input methods for reviewing and correcting outcomes in object detection and activity recognition in videos. Our findings indicate that while some complex input methods, such as free-text, require more time, the quality and impact of their feedback on model accuracy often surpass those of simpler methods that require less effort. However, more effort does not always lead to better-quality feedback, especially when aiming to improve the model. Our VLM experiments show that the most accurate models were trained using detailed natural language feedback or precise word-level corrections, while simple yes/no judgments also led to solid performance at a much lower annotation cost.
- Open Access
- 25 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 58 (9) , 1-37
A Systematic Review on Human Roles, Solutions, and Methodological Approaches to Address Bias in AI
People play a significant role in designing, developing, and employing artificial intelligence (AI) systems. They can consider contextual information beyond the scope of AI models, thereby influencing system outcomes. At the same time, people’s choices or biases can introduce problems into the systems. This paradoxical scenario, in which people can both introduce and contribute to relieving the inherited machine bias, demands comprehensive and multidisciplinary approaches involving informed human interventions to improve systems’ performances and reduce their biases. Researchers across various communities have investigated multifaceted methods to reduce and mitigate bias in AI systems. Regardless of the method, humans are always involved in the debiasing method in one way or another, emphasizing the importance of human intervention during AI systems development. In this systematic review, we analyzed 100 peer-reviewed publications from various human-computer interaction (HCI) and machine learning (ML) venues. We discuss their research efforts to minimize data bias and algorithmic bias from three angles. First, we present a comprehensive taxonomy of bias mitigation solutions, analyzing the research methodologies and standard benchmarks for evaluating these solutions, highlighting the human researcher’s role in developing and evaluating solutions to address bias. Next, we identify humans’ roles in alleviating biases and specify how, when, and where their involvement occurs within the AI lifecycle. Finally, we summarize the research focus and methodologies across research disciplines. Our review revealed that, while technical solutions are essential, addressing bias requires a broad perspective that integrates human oversight, ethical frameworks, and interdisciplinary collaboration.
- 25 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 58 (9) , 1-37
Open AccessA Systematic Review on Human Roles, Solutions, and Methodological Approaches to Address Bias in AI
People play a significant role in designing, developing, and employing artificial intelligence (AI) systems. They can consider contextual information beyond the scope of AI models, thereby influencing system outcomes. At the same time, people’s choices or biases can introduce problems into the systems. This paradoxical scenario, in which people can both introduce and contribute to relieving the inherited machine bias, demands comprehensive and multidisciplinary approaches involving informed human interventions to improve systems’ performances and reduce their biases. Researchers across various communities have investigated multifaceted methods to reduce and mitigate bias in AI systems. Regardless of the method, humans are always involved in the debiasing method in one way or another, emphasizing the importance of human intervention during AI systems development. In this systematic review, we analyzed 100 peer-reviewed publications from various human-computer interaction (HCI) and machine learning (ML) venues. We discuss their research efforts to minimize data bias and algorithmic bias from three angles. First, we present a comprehensive taxonomy of bias mitigation solutions, analyzing the research methodologies and standard benchmarks for evaluating these solutions, highlighting the human researcher’s role in developing and evaluating solutions to address bias. Next, we identify humans’ roles in alleviating biases and specify how, when, and where their involvement occurs within the AI lifecycle. Finally, we summarize the research focus and methodologies across research disciplines. Our review revealed that, while technical solutions are essential, addressing bias requires a broad perspective that integrates human oversight, ethical frameworks, and interdisciplinary collaboration.
- 20 November 2025
- Published by Elsevier in International Journal of Human-Computer Studies
- Vol. 208
User judgment of an AI model is biased by its description: A study in a job interview training context
- 20 November 2025
- Published by Elsevier in International Journal of Human-Computer Studies
- Vol. 208
User judgment of an AI model is biased by its description: A study in a job interview training context