Acceptable Use of AI in an SBRI Application Process
1. Purpose and scope
This document outlines SBRI Healthcare’s stance on using generative AI during the funding application and evaluation stages. It details the specific obligations for everyone involved in these processes.
As stewards of public funds, our priority is to maintain top-tier research standards by supporting the most impactful innovations. We aim to embrace the advantages of generative AI while carefully balancing its potential downsides.
2. Contents
- Purpose & scope
- Contents
- Principles
- Policy on Permissible AI Use
4.1 Applicant Responsibilities
4.2 Assessor Responsibilities
4.3 SBRI Healthcare Responsibilities - Enforcing the Policy
5.1 During the Application Phase
5.2 During the Assessment Phase - Terminology
- Related Documentation
3. Principles
A trustworthy funding and review process is the backbone of our system, rooted in transparency and integrity. The UK’s global leadership in innovation stems from the human capacity for original thought and creative curiosity. Protecting these values ensures a healthy research ecosystem that serves the public good. Maintaining this integrity is a shared responsibility. This policy clarifies the boundaries for using generative AI for both our organization and our wider community.
4. Policy on permissible AI use
4.1 Applicant responsibilities
This section defines the expectations for those drafting funding proposals.
Data Security: Exercise extreme caution when using AI tools. Never input personal or sensitive details without explicit, formal permission from the data owner.
Quality Control: Be alert to potential biases in AI-generated content and take steps to correct them.
Integrity: Ensure AI-produced material is free from plagiarism, fabrications, or unauthorised confidential data. All submissions must adhere to current data protection and intellectual property laws.
Transparency: In line with industry standards, applicants should disclose if AI was used during the drafting process. Such a disclosure will not negatively impact the final score.
Interviews: The use of generative AI is strictly prohibited during live interviews that form part of the application and assessment process.
4.2 Assessor responsibilities
We value our reviewers for their unique professional insights and expert judgment. Consequently, assessors must:
Human-Only Review: Avoid using generative AI for any part of the evaluation, including basic tasks like proofreading or formatting.
Legal Compliance: Follow all relevant IP and data privacy regulations.
Impartiality: Base scores solely on the merit of the application; do not speculate on whether the applicant used AI.
Confidentiality: Not upload, share, or otherwise input proprietary or confidential application materials into any generative AI or third-party tools. Doing so constitutes a breach of trust and policy.
Accessibility: If assistive tech is required for accessibility, assessors must use nongenerative tools that process information locally rather than via the cloud.
4.3 SBRI Healthcare responsibilities as a funder
We commit to providing transparent, easy-to-find guidance so all participants understand our expectations regarding AI.
Human-Only Funding Decisions: SBRI Healthcare uses AI-powered tools to help summarise feedback provided by external reviewers which is then checked by team members before being shared with applicants. However, it is important to note, the shortlisting and funding decisions are made solely by our assessors and Panel, with no AI involvement in the outcome determination.
5. Enforcing the Policy
5.1 During the Application Phase
If a claim of research misconduct involving AI is substantiated, SBRI Healthcare may take the following actions:
• Disqualify any active applications involving the individual.
• Ban the individual from future funding opportunities, potentially permanently.
• Increase monitoring of the individual’s existing projects.
• Mandate the individual's removal from a funded team.
• Revoke or reclaim funding from the host organisation related to that individual’s work.
• Bar the individual from serving on future review panels or advisory boards.
5.2 During the Assessment Phase
If an assessor informs us that they have used generative AI:
• Reviewers: Their written feedback will be discarded.
• Panellists: Their input will be excluded from final funding deliberations.
• Transparency: The affected applicant will be notified of the breach and the steps
taken to ensure a fair, human-led re-evaluation.
6. Terminology
There are no special terms or definitions used in this Policy.
7. Related documents
List of related documents.
SBRI Healthcare Privacy Policy
SBRI Confidentiality Guidance