Request for Proposals

Scientific Scope

Data science and AI methodology for research. Develop novel data science and AI methodologies that are explicitly designed to advance scientific and scholarly research. Projects should address methodological gaps that arise in real research contexts and aim to generate models, algorithms, or methodological paradigms that go beyond serving as a one-off application to a specific dataset. Examples include, but are not limited to:

• Hybrid AI systems that combine symbolic reasoning, causal modeling, or knowledge graphs with neural networks.
• Data science methods for harmonizing research data collected across institutions, disciplines, or sectors.
• Methods for validating and benchmarking AI systems used in research contexts.
• AI models that support interpretive scholarship while preserving nuance and uncertainty.

Translating methodologies to tools. Design, implement, and evaluate AI tools, AI-enabled research workflows, guardrails, and shared resources that translate methodological advances into robust, reproducible, and scalable research practice. Projects should move beyond methodology development to practical research solutions, including human-AI collaboration, validation frameworks, and generalizable tools that serve defined research communities. Proposals should identify collaborators who will serve as early adopters of the tools and workflows, and propose how they will be sustained, shared, or scaled. Examples include, but are not limited to:

• AI tools for a workflow for systematic review, literature synthesis, or meta-analysis that integrate human verification.
• Systems for tracking AI model versions, prompts, fine-tuning steps, and training data used in research analyses.
• Domain-specific AI toolkits that can be adopted by multiple labs
• Modular, open-source workflows that enable researchers to plug in new models while preserving comparability across projects.

AI-enabled expansion of the research frontier. Use AI to enable a fundamentally new class of research questions that were previously infeasible to ask or answer. Projects should demonstrate how AI makes possible new forms of observation, inference, integration, or exploration. Proposals must clearly articulate why the research question was previously out of reach, what specific technical or conceptual barrier AI overcomes, and how the new research would advance the field. Projects that merely apply AI to address an established question more efficiently, at larger scale, or with improved predictive performance will not be competitive. Examples include, but are not limited to:

• Reconstructing historical or cultural processes from fragmented archival records at scale.
• Using generative or simulation-based AI systems to test theoretical scenarios that cannot be experimentally realized.
• Discovering latent structures that challenge established disciplinary assumptions.
• Identifying new units of analysis that were previously undetectable.

Emerging research directions and community formation. Establish and operationalize a strategically positioned research community around an emerging and high-leverage research direction that is data- and / or AI-intensive. Projects should define a focused intellectual agenda, build collaboration, and create shared assets that position U-M to lead in this direction. Proposals must articulate why the direction is strategically significant and time-sensitive, what barriers will be overcome, and how the effort will produce concrete outputs and enable major external funding opportunities. Examples drawn from previous or current efforts: AI and quantum, single cell-genomics, sequential decision making for clinical applications, etc.