Nine faculty members from Carnegie Mellon University’s School of Computer Science recently received Google Academic Research Awards , which aim to fund and actively collaborate with researchers to generate meaningful work with real-world applications. Award amounts vary but can provide up to $100,000 for their duration.
SCS faculty received awards to make education equitable, accessible and effective using artificial intelligence; apply Google tools like Gemini to systems and infrastructure problems; and use AI to benefit society.
Paulo Carvalho , an assistant professor in the Human-Computer Interaction Institute (HCII), and Aniket Kittur , a professor in the HCII, received an award for their project, "From Engagement to Learning: Augmenting Videos for Skill Learning With LLMs." Their work seeks to improve how people learn from videos. Carvalho and Kittur plan to develop a system to find multiple high-quality videos on a topic, extract key skills, bug cluster them into meaningful minilessons, and generate an interactive quiz-like overview to help learners decide which videos to watch and prime them to learn more. The system will keep people’s attention, increase engagement with materials, and lead to greater learning and interest without editing or creating videos.
Sherry Wu , an HCII assistant professor, and Ken Koedinger , the Hillman Professor and director of the Master of Educational Technology and Applied Learning Sciences program in the HCII, received an award for their project, "Explicitly Train Humans Toward Human-AI Collaboration." Human-AI collaboration will become increasingly important as large language models (LLMs) perform more tasks. Wu and Koedinger’s work aims to train students to better use, understand and trust AI. They will integrate human-AI interaction training into curricula and create an LLM tutoring platform that can accommodate varying difficulties of tasks, different learning objectives, and student weaknesses to help prepare students for an AI-integrated workforce.
Zhihao Jia , an assistant professor in the Computer Science Department , received an award for his project, "An ML-Optimized System for ML" to help build fast, scalable and secure machine learning (ML) systems. The work relies on new techniques to automate the generation, verification and application of machine learning optimizations for different tasks, deploying these optimizations in an end-to-end fashion. His team’s techniques will enable systems and scientists to speed up their research and operational workloads at reduced costs using distributed computing. This will significantly lower the barrier of developing and using modern machine learning techniques, allowing everyone to benefit from AI advances.
Ken Holstein , an assistant professor in the HCII, and Haiyi Zhu , the Daniel P. Siewiorek Associate Professor in the HCII, received an award for their project, "Participatory AI Measurement Design: Supporting Early Stakeholder Engagement in the Design of AI Measurements." AI applications often fail because developers make faulty assumptions when designing concrete measures for abstract concepts like worker wellbeing, risk and fairness. To address this, Holstein and Zhu will develop a new methodology and interactive toolkit for participatory measurement design for AI to bring together AI developers, domain experts and people who may be affected by the tools.
Maarten Sap , an assistant professor in the Language Technologies Institute , received an award for his project, "PARTICIP-AI: Studying Lay People’s Needs, Judgments and Impact Assessment for Future AI Use Cases and AI Dilemmas." The expansion of AI and LLMs has led to increased risks of misuse and under-addressed AI needs specific to diverse communities. For AI development to benefit society holistically, it should reflect the perspectives of various communities. Sap and his team propose PARTICIP-AI, a two-step framework that first gathers future AI needs and dilemmas through speculative, design-based, brainstorming surveys and then collects diverse community opinions on the acceptability and potential impacts of contentious AI use cases.
Motahhare Eslami , an assistant professor in the HCII and the Software and Societal Systems Department , received an award with Kayhan Batmanghelich of Boston University for their project, "Bias Awareness and Mitigation in Breast Cancer Risk Prediction for Marginalized Groups Through Participatory AI." While current AI-powered breast cancer risk-assessment models are crucial for early detection, they are primarily tailored for non-Hispanic, white women and do not perform well for women of color, particularly Black women. Black women, however, have a significantly higher breast cancer mortality rate than non-Hispanic women. The project will integrate bias awareness and mitigation into AI models for predicting breast cancer risk for marginalized groups, particularly Black women, through collaboration with clinicians, community hospitals and marginalized stakeholders.
Learn more about the awards on the Google Academic Research Award program website.