MIT Generative AI Impact Consortium Seed Grants

The MIT Generative AI Impact Consortium (MGAIC) announced a call for proposals for innovative seed projects to study high impact uses of generative AI models. Projects are intended to be interdisciplinary, collaborative, and open source, with a focus on several key areas of AI impact, including, but not limited to: design, education, engineering, healthcare, sustainability, and workforce transformation. The inaugural call for proposals ended March 7, 2025.

Call for proposal details

Projects

The projects are divided into four funding classes:

  • Exploratory projects: up to $50,000 focused for example individual PI work
  • Innovation projects: up to $150,000 with student/postdoc focused engagement with one or more PIs
  • Flagship projects: up to $300,000 with multiple students/postdocs and at least two PIs
  • Company-directed projects: member companies have the opportunity to directly fund selected projects from above or make special arrangements with the PIs.

  • “A Pilot Study on LLM-Based Extension Services in Smallholder Agriculture,” led by Ali Aouad (MIT Sloan);
  • “AI Adoption and Organizational Inequality,” led by Nathan Wilmers (MIT Sloan);
  • “The Impact of US Electricity Infrastructure on Data Center Development,” led by Mert Dermier (MIT Sloan);
  • “HypSynth: Generative AI for Science through Multi-Agent Hypothesis Refinement,” led by Satrajit Ghosh (McGovern Institute for Brain Research);
  • “Leveraging Large Language Models to Incorporate Qualitative Information into Decision-Making Models in Transportation and Logistics,” led by Alexandre Jacquillat (MIT Sloan);
  • “Pareto Machines: Training AI Agents to Find Positive-Sum Solutions,” led by Sinan Aral (MIT Sloan);
  • “Physics Predictions from Physical Generative AI,” led by Jesse Thaler (Physics).

  • “3D Generative AI for Physical Realism: Encoding Physical Properties into Generative AI models to Create Physically Valid 3D Models,” led by Stefanie Mueller (EECS; MechE);
  • “Advancing the Potential of In-Silico Trials and Generative AI in Health with Controllable Diffusion Models and Global Outreach,” led by Elazer Edelman (IMES) and Joseph J. Frassica (IMES);
  • “Agentic AI for Interpretative and Predictive Bio-inspired Materials Discovery Using Autonomous Experimentation and Self-Learning,” led by Markus J. Buehler (CEE);
  • “AI Coach for Skilled Career Navigation,” led by John Hart (MechE) and John Liu (MechE);
  • “AI Empowered Human-Centered Design: From Identifying Latent Human Needs to Generating Human-Guided Designs of Physical Systems,” led by Caitlin Mueller (CEE; Architecture) and Maria Yang (MechE);
  • “Black-Box Auditing with LLM Simulated Agents,” led by Chara Podimata (MIT Sloan);
  • “Building a Time Series Foundation Model to Listen to the Universe,” led by Philip Harris (Physics), Erik Katsavounidis (Physics), Yury Polyanskiy (EECS), George Ricker (MIT Kavli Institute), and Daniela Rus (CSAIL);
  • “AI-Driven Tutors and Open Datasets for Early Literacy Education,” led by Satrajit Ghosh (McGovern Institute for Brain Research) and John Gabrieli (McGovern Institute for Brain Research);
  • “Bridging Statistical Physics and Generative AI: Scaling Laws, Robustness, and Spectral Control,” led by Marin Soljacic (Physics);
  • “Conditional Generation of Micro-Level Household Consumption with Applications to Energy,” led by Christopher Knittel (MIT SLOAN);
  • “Developing jam_bots: Real-Time Collaborative Agents for Live Human-AI Musical Improvisation,” led by Joseph A. Paradiso (MIT Media Lab), Anna Huang (Music and Theater Arts; EECS), and Eran Ergozy (Music and Theater Arts);
  • “Efficient GPU Sparse Automatic Differentiation for Scientific Computing,” led by Justin Solomon (EECS) and Jonathan Ragan-Kelley (EECS);
  • “Empowering Underserved Students: AI-Driven Calculus Tutoring for Equitable Education,” led by Eric Klopfer (Comparative Media Studies/Writing) and Cynthia Breazeal (MIT Media Lab);
  • “Enhancing Biodiversity Image Datasets with Generative AI for Improved Species Classification,” led by Sara Beery (EECS) and Fabio Duarte (DUSP);
  • “Generative AI as a Therapeutic Intervention for Alleviating Insistence on Sameness in Autism,” led Pawan Sinha (Brain and Cognitive Sciences);
  • “Generative AI for Automated Experimental Setup and Alignment in Optics and Beyond,” led by Marin Soljacic (Physics) and Pulkit Agrawal (EECS);
  • “Generative AI for Predicting Citrus Greening Spread,” led by Sherrie Wang (MechE; IDSS);
  • “Generative AI for Enhancing Risk Management of Software Supply Chain,” led by Retsef Levi (MIT Sloan);
  • “GENIUS: GENerative Intelligence for Urban Sustainability,” led by John E. Fernández (Architecture), Omar Khattab (EECS), and Norhan Bayomi (MIT Environmental Solutions Initiative);
  • “Glia: An AI Assistant to Design High-Performance GenAI Systems,” led by Mohammad Alizadeh (EECS) and Hari Balakrishnan (EECS);
  • “Giving Power Back to Humans in Visual Generative AI though Modularity and Compositionality,” led by Fredo Durand (EECS);
  • “Leveraging LLM dialogues to Build Trust in Critical Institutions,” led by Adam Berinksy (Political Science) and David Rand (MIT Sloan; Brain and Cognitive Sciences);
  • “Measuring and Mitigating Homogenization in Generative AI,” led by Manish Raghavan (MIT Sloan; EECS);
  • “Multimodal Exploratory Data Analysis with Semiformal Programming,” led by Arvind Satyanarayan (EECS);
  • “Multimodal Generative AI Framework for Brain Age Prediction and Future MRI Forecasting,” led by Laura Lewis (EECS; IMES) and Marzyeh Ghassemi (EECS; IMES);
  • “Neuromechanics-Grounded Imitation Learning for Predicting Locomotor Stability,” led by Nidhi Seethapathi (Brain and Cognitive Sciences);
  • “P3: A Generative AI Foundation Model for PDE Surrogate Modeling and Discovery,” led by Sili Deng (MechE);
  • “Personalized Learning with GenAI for MIT Open Learning Courseware,” led by Dimitris Bertsimas (MIT Sloan) and Ana Trisovic (EECS);
  • “Random Utility Models Meet LLM Alignment,” led by Vivek Farias (MIT Sloan) and Ali Aouad (MIT Sloan);
  • “Spectrum-Conditioned Generative Elucidation of Unknown Molecules,” led by Connor Coley (ChemE; EECS);
  • “Understanding Human Music Perception Using Generative AI,” led by Josh McDermott (Brain and Cognitive Sciences) and Anna Huang (Music and Theater Arts; EECS);
  • “Voices of the Poor,” led by Sendhil Mullainathan (EECS; Economics) and Ashesh Rambachan (Economics);
  • “Zero-Waste Tree: Neural Radiance Field Segmentation for Generative Timber Design,” led by Caitlin Mueller (CEE; Architecture) and Skylar Tibbits (Architecture).

  • “A2rchi – AI Support for Classes and Research Resources Teams,” selected by TWG, and led by Tim Kraska (CSAIL) and Christoph Paus (Physics);
  • “Building Robust and Scalable Enterprise Data Curation Applications with Agentic AI,” selected by TWG, and led by Samuel Madden (EECS);
  • “Efficient Robot Learning via Multimodal Feedback and VLM-Guided Augmentation,” selected by Tata, and led by Andreea Bobu (Aeronautics and Astronautics);
  • “High-Resolution and Multi-Modality Sense of Touch: Compact Terahertz Tactile Chip System for Advanced Robotic Training and Manipulation,” selected by Analog Devices, and led by Ruonan Han (EECS);
  • “Manufacturability-Aware Generative AI for Next-Gen Topology Optimization,” selected by Tata, and led by Faez Ahmed (MechE) and Josephine C. Carstensen (CEE);
  • “Multi-Agent Large Language Models for Battery and Semiconductor Applications,” selected by SK Telecom, and led by Ju Li (Nuclear Science and Engineering), Kaiming He (EECS), and Rafael Gómez-Bombarelli (Materials Science and Engineering);
  • “Multimodal Tactile Sensing for Robotics,” selected by Analog Devices, and led by Wojciech Matusik (EECS) and Paul Liang (MIT Media Lab; EECS);
  • “Understanding How LLM Agents Deviate from Human Choices,” selected by SK Telecom, and led by Pattie Maes (MIT Media Lab).