Our Use of Artificial Intelligence
The Thrasher Research Fund is exploring the responsible use of artificial intelligence (AI) to strengthen our grant review process. We believe AI has the potential to complement human expertise by helping us review applications more thoroughly, identify relevant scientific information, and provide more useful feedback to researchers.
Protecting applicant information
Protecting confidential application information is a priority. We use only enterprise-level AI services from OpenAI and Anthropic. Under these services, your application is never used to train AI models, data is encrypted in transit and at rest, and both providers meet recognized enterprise security standards, including SOC 2 Type II.
How we think about AI
We have invested substantial effort in building a structured framework for AI-assisted review. Based on our grantmaking criteria and staff experience, we developed detailed rubrics, decision rules, and instructions that guide how AI assesses proposals. Our goal is for AI to apply our standards as our staff would, and in some respects to assess applications more thoroughly than staff alone can, by drawing on a broader range of scientific literature.
We recognize that AI can make errors, overlook important context, or produce inaccurate information. These limitations are central to our testing and evaluation.
Concept paper review
Thrasher Research Fund staff review every concept paper and make all decisions on which concept papers move forward to full proposal. Independently from staff reading and scoring, we are testing an AI to review and evaluate concept papers. We will compare the AI assessments with staff assessments and carefully examine areas where they differ. Human judgment will be responsible for all final decisions.
Identifying peer reviewers
For E.W. "Al" Thrasher full proposals, AI will suggest potential peer reviewers. This will include identifying potential conflicts of interest, including recent publications or collaborations involving investigators. Thrasher Research Fund staff evaluate potential reviewers and make all reviewer selections.
Why we are exploring AI
Our funding programs cover a broad range of child health research, and no individual reviewer or staff member can have deep expertise in every scientific area we encounter.
We anticipate that AI will:
- identify relevant literature, clinical trials, and related research that could otherwise be missed
- recognize important scientific issues earlier in the review process
- provide applicants with more substantive and useful feedback than staff can provide on a large batch of applications
- over time, make portions of the review process more efficient while maintaining appropriate human oversight
Questions or concerns about AI use
Applicants with concerns about the use of AI in reviewing their concept paper may contact a Thrasher Research Fund Research Manager to discuss them.