Task Force Meeting 7
- June Burke
Meeting Summary
The team discussed various aspects of utilizing AI tools, especially large language models (LLMs), for managing research and academic work. The topics covered include the use of LLMs for tasks like grant pre-review, document management, research data analysis, and the ethical and legal implications of AI in these areas.
Grant Pre-Review Proposal: The team explored the idea of using a local LLM for a grant pre-review process. They discussed the benefits of using AI tools to assist in reviewing research proposals and the potential for automating the workflow. A proposal and budget for this initiative are to be developed, with a focus on the capabilities of the AI tool to streamline the pre-review process.
IRB and Ethical Considerations: Ethical concerns were raised regarding the use of AI tools for research, particularly from an IRB (Institutional Review Board) perspective. There were discussions on ensuring compliance with ethical guidelines while using AI to analyze sensitive or proprietary research data.
AI and Data Management: The conversation highlighted the opportunities and challenges associated with using AI to manage complex research datasets. While AI could assist in organizing and interpreting large datasets, the potential for violating intellectual property or data security was a significant concern. The discussion emphasized the need to assess how proprietary data is handled when using AI tools like ChatGPT, especially in relation to intellectual property rights and security.
Proprietary Data and Intellectual Property: The issue of AI tools, such as OpenAI’s models, accessing and potentially leaking proprietary or sensitive data was raised. Examples were shared of AI systems analyzing documents or internal data without authorization, leading to concerns about data security and the protection of intellectual property.
AI’s Growing Role and Challenges: While there is a potential for AI to improve access to research documents and datasets within an organization, there are significant challenges around ensuring privacy and confidentiality. The team discussed the importance of developing clearer guidelines and policies on when and how to use AI tools for managing research data, particularly for sensitive documents.
Privacy, Copyright, and Accessibility: There was a focus on the importance of maintaining data privacy and protecting copyright in research documents when using AI tools. The team differentiated between public research data and sensitive internal documents, noting that while federally funded research may be more accessible, internal data, such as organizational policies and procedures, must be safeguarded.
Data Protection and Compliance: Concerns were raised about data protection, particularly in the context of using AI systems that may not always adhere to confidentiality protocols. The team discussed working with tech transfer offices to evaluate and mitigate the risks associated with the use of these tools.
Policy and Standards for AI in Research: The need for clearer standards and protocols on AI usage was highlighted. A potential collaboration with tech transfer offices was suggested to better understand the risks and regulatory compliance needed when incorporating AI into research workflows.
Key Takeaways:
Grant Pre-Review and Proposal Development: The team is moving forward with the idea of using a local LLM for grant pre-reviews, and a detailed proposal and budget will be created for this initiative.
Ethical Considerations and IRB: Ethical concerns, particularly regarding data privacy and compliance, need to be addressed as AI is integrated into research workflows. The team is considering how the IRB will assess AI usage in these contexts.
AI’s Role in Data Management: AI could significantly enhance the management and analysis of complex research datasets, but risks related to intellectual property and data security must be carefully considered.
Proprietary Data Security: There are concerns about AI tools (like OpenAI’s models) accessing and potentially exposing proprietary or sensitive data. The team is cautious about using AI for proprietary research unless data protection mechanisms are firmly in place.
Balancing Privacy and Accessibility: AI tools have the potential to improve access to documents and research data, but ensuring that sensitive or proprietary information is protected remains a significant challenge.
Data Protection Protocols: The team is considering working with tech transfer offices to better understand the risks and legal aspects of using AI tools for analyzing research data, ensuring compliance with intellectual property laws and privacy standards.
Clarifying Policies for AI in Research: Clear policies and guidelines for AI’s role in research are essential, particularly in managing the risks associated with the use of AI in sensitive areas like data analysis, document management, and research pre-reviews.
This consolidated summary provides an overview of the key points discussed and the action items to be addressed going forward.