Mastodawn

Harvard HCI May 15, 2025

Krzysztof Gajos May 15, 2025

@zbucinca's latest #chi2025 paper shows that people who receive AI decision recommendations supported by contrastive explanations (choose A instead of B because..) help people grow their skills. But this only happens if the alternative (the B in the contrastive explanations) is something that people would plausibly consider. This important because earlier work showed that people do not learn when AI provides conventional explanations (reasons for/against a decision).

https://iis.seas.harvard.edu/papers/bucinca2025contrastive.pdf

Harvard HCI May 15, 2025

Krzysztof Gajos May 15, 2025

Congratulations to @zm003 who has recently defended his PhD dissertation! Over the past 6 years, Zilin focused on understanding and designing technology that benefits vulnerable groups. He analyzed the benefits and perils of LLM chatbots in supporting well-being of LGBTQ+ people; he demonstrated how dating web sites could be redesigned to reduce implicit racial bias when people choose prospective partners; and he developed tools to support front line negotiators.

Harvard HCI Apr 3, 2025

Krzysztof Gajos Apr 3, 2025

Congratulations to (soon to be) Dr. Zana Buçinca (@zbucinca) for defending her dissertation yesterday!

In her PhD, Zana demonstrated that human cognitive engagement moderates the effectiveness of AI support in human decision making, she introduced cognitive forcing functions, and has launched the new sub-field of worker-centric AI.

Her upcoming #CHI2025 paper on Contrastive Explanations That Anticipate Human Misconceptions exemplifies this latest direction in her work.

https://iis.seas.harvard.edu/papers/bucinca2025contrastive.pdf

Harvard HCI Oct 9, 2024

Elena Glassman Oct 3, 2024

Come be my colleague!😀 https://academicpositions.harvard.edu/postings/14166

Tenure-Track Professor in Computer Science

The Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) seeks applicants for a tenure-track position in Computer Science, with an expected start date of July 1, 2025. We invite applications from all areas of Computer Science.Computer Science at Harvard benefits from outstanding undergraduate and graduate students, world-leading faculty, significant industrial collaboration, and substantial support from SEAS. Information about Harvard’s current faculty, research, and educational programs in computer science is available at http://www.seas.harvard.edu/computer-science.SEAS celebrates the multiple dimensions of diversity that each member of our community offers, including diversity of background, perspective, and lived experience. We strongly welcome applications from persons from underrepresented groups.

Harvard HCI Apr 1, 2024

Elena Glassman Mar 29, 2024

This was fun---a chat with the student newspaper: https://www.thecrimson.com/article/2024/3/29/elena-glassman-15q/

Fifteen Questions: Elena Glassman on Human-Computer Interaction, Freestyle Wrestling, and MIT Dorms | Magazine | The Harvard Crimson

The human-computer interaction expert sat down with FM to discuss software design, the importance of bicycling infrastructure, and her time competing in women’s freestyle wrestling.

Harvard HCI Apr 1, 2024

Zana Buçinca Mar 12, 2024

AI assistance is impacting our decisions and the quality of our work. But how will this assistance affect us -- our skills, our growth and improvement, enjoyment, collaboration, or our agency in the workplace? The current design of AI assistance does not consider human-centric objectives; we need methods to account for them.

We propose offline RL as an approach for optimizing such human-centric objectives in AI-assisted decision-making.

Link to preprint: http://arxiv.org/abs/2403.05911

Towards Optimizing Human-Centric Objectives in AI-Assisted Decision-Making With Offline Reinforcement Learning

As AI assistance is increasingly infused into decision-making processes, we may seek to optimize human-centric objectives beyond decision accuracy, such as skill improvement or task enjoyment of individuals interacting with these systems. With this aspiration in mind, we propose offline reinforcement learning (RL) as a general approach for modeling human-AI decision-making to optimize such human-centric objectives. Our approach seeks to optimize different objectives by adaptively providing decision support to humans -- the right type of assistance, to the right person, at the right time. We instantiate our approach with two objectives: human-AI accuracy on the decision-making task and human learning about the task, and learn policies that optimize these two objectives from previous human-AI interaction data. We compare the optimized policies against various baselines in AI-assisted decision-making. Across two experiments (N = 316 and N = 964), our results consistently demonstrate that people interacting with policies optimized for accuracy achieve significantly better accuracy -- and even human-AI complementarity -- compared to those interacting with any other type of AI support. Our results further indicate that human learning is more difficult to optimize than accuracy, with participants who interacted with learning-optimized policies showing significant learning improvement only at times. Our research (1) demonstrates offline RL to be a promising approach to model dynamics of human-AI decision-making, leading to policies that may optimize various human-centric objectives and provide novel insights about the AI-assisted decision-making space, and (2) emphasizes the importance of considering human-centric objectives beyond decision accuracy in AI-assisted decision-making, while also opening up the novel research challenge of optimizing such objectives.

arXiv.org

Harvard HCI Feb 23, 2024

Zilin Ma Feb 20, 2024

Would you come out to ChatGPT? Or should you? Check out our paper (w/Yiyang Mei, Yinru Long, Nick Su, @kgajos)where we studied how LGBTQ+ people used LLM-based chatbots for mental health support. https://arxiv.org/abs/2402.09260 #CHI2024 #LGBTQ #mentalhealth #LLM #ChatGPT

Evaluating the Experience of LGBTQ+ People Using Large Language Model Based Chatbots for Mental Health Support

LGBTQ+ individuals are increasingly turning to chatbots powered by large language models (LLMs) to meet their mental health needs. However, little research has explored whether these chatbots can adequately and safely provide tailored support for this demographic. We interviewed 18 LGBTQ+ and 13 non-LGBTQ+ participants about their experiences with LLM-based chatbots for mental health needs. LGBTQ+ participants relied on these chatbots for mental health support, likely due to an absence of support in real life. Notably, while LLMs offer prompt support, they frequently fall short in grasping the nuances of LGBTQ-specific challenges. Although fine-tuning LLMs to address LGBTQ+ needs can be a step in the right direction, it isn't the panacea. The deeper issue is entrenched in societal discrimination. Consequently, we call on future researchers and designers to look beyond mere technical refinements and advocate for holistic strategies that confront and counteract the societal biases burdening the LGBTQ+ community.

arXiv.org

Harvard HCI Oct 6, 2023

Curtis Wilcox Oct 5, 2023

Digital Accessibility Services (DAS) at Harvard University is looking for a Digital Accessibility Specialist to help grow our team's capacity to support the Harvard community make its digital presence more accessible to everyone.

If you are interested or know someone who might be, please apply and/or send them the link! We have a great team culture, a supportive employer, and get to do meaningful work in a dynamic environment. Come join us!

https://sjobs.brassring.com/TGnewUI/Search/Home/Home?partnerid=25240&siteid=5341#jobDetails=1998039_5341

#accessibility #a11y #jobs #hiring #HigherEd

Job Search Welcome | Harvard University

Visit Harvard Careers to explore job opportunities, create a profile, and apply for open positions.

Harvard HCI Sep 28, 2023

Ian Arawjo Sep 19, 2023

A pre-print of our paper, “ChainForge: A Visual Toolkit for Prompt Engineering and LLM Hypothesis Testing,” is now up on arXiv. This is research with Elena Glassman, Martin Wattenberg, Chelse Swoopes and Priyan Vaithilingam: https://arxiv.org/abs/2309.09128

ChainForge: A Visual Toolkit for Prompt Engineering and LLM Hypothesis Testing

Evaluating outputs of large language models (LLMs) is challenging, requiring making -- and making sense of -- many responses. Yet tools that go beyond basic prompting tend to require knowledge of programming APIs, focus on narrow domains, or are closed-source. We present ChainForge, an open-source visual toolkit for prompt engineering and on-demand hypothesis testing of text generation LLMs. ChainForge provides a graphical interface for comparison of responses across models and prompt variations. Our system was designed to support three tasks: model selection, prompt template design, and hypothesis testing (e.g., auditing). We released ChainForge early in its development and iterated on its design with academics and online users. Through in-lab and interview studies, we find that a range of people could use ChainForge to investigate hypotheses that matter to them, including in real-world settings. We identify three modes of prompt engineering and LLM hypothesis testing: opportunistic exploration, limited evaluation, and iterative refinement.

arXiv.org

Harvard HCI Aug 25, 2023

Prof. Elena Glassman (@eglassman) will be one of the keynote speakers at #vlhcc23 in October in Washington DC. Come and hear about Human Inspection of Code at Scale: The Value of Variation in Informing Decision-Making!

https://conf.researchr.org/details/vlhcc-2023/vlhcc-2023-keynotes/2/Human-Inspection-of-Code-at-Scale-The-Value-of-Variation-in-Informing-Decision-Makin

Human Inspection of Code at Scale: The Value of Variation in Informing Decision-Making (VL/HCC 2023 - Keynotes) - VL/HCC 2023

From the beginning of the computer age, people have sought easier ways to learn, express, and understand computational ideas. Whether this meant moving from punch cards to textual languages, or command lines to graphical UIs, the quest to make computation easier to express, manipulate, and understand by a broader group of people is an ongoing challenge. The IEEE Symposium on Visual Languages and Human-Centric Computing is the premier international forum for research on this topic. Established in 1984, the mission of the conference is to support the design, theory, application, and evaluat ...