AI Research Day
Event Planning:
Discussed plans for an AI Research Day tentatively scheduled for October 22nd. The event aims to explore the integration of AI into public health research.
Keynote Speaker: Tentatively confirmed Ziad Obermeyer as the keynote speaker. Plans for a substantive half-day event were discussed.
Posters and Awards: Planned for poster presentations, with awards for the best abstracts. A selection committee will be formed to review submissions.
Breakout Sessions: Ideas were proposed for breakout sessions to facilitate direct interactions among attendees. Topics included ethical use of AI, industry-academia partnerships, and responsible conduct of research.
Format: Consideration was given to the event format, including whether to host the breakout sessions on the same day or spread them over multiple days.
Additional Activities:
Corporate Partnerships: Discussion on holding a separate event focused on industry partnerships and collaborations.
Responsible Conduct of Research: Plans for a seminar on incorporating AI ethically and responsibly into research, particularly for trainees.
Timing and Logistics:
Consideration was given to the timing of the event and related activities, including the feasibility of hosting multiple events and ensuring attendee engagement.
Next Steps:
Further discussion and planning needed to finalize event details, including session topics, speakers, and logistics.
Action Items:
Determine final event details, including session topics and speakers.
Confirm keynote speaker and poster presentation guidelines.
Coordinate with relevant departments and stakeholders for fall seminar planning and promotion. Once topic is defined, we can help support with a speaker and date.
Proposal for AI Production Office
The meeting also consisted of on discussions regarding setting up an AI production office within the School of Public Health. Ahmed Hassoon outlined the proposal, emphasizing the need for a secure server at the Biostatistics department to host open-source foundation models. The office would offer support services to researchers who want to leverage AI-related tools but lack the expertise or resources to do so independently.
Key points of the proposal included:
Purpose: To provide researchers with access to AI tools and resources, facilitating collaboration and enhancing education capabilities.
Scope: Initially focusing on language model hosting, with potential for custom model development and consultancy services.
Implementation Phases: Planning and infrastructure setup, followed by a demonstration project to showcase capabilities, with continuous improvement and scaling.
Budget: A modest budget request, including funding for GPUs and backup power supplies, with optional additional costs for demonstration projects.
The proposal aimed to streamline the integration of AI into research projects, enabling researchers to utilize AI tools effectively without needing extensive expertise in AI programming. The meeting also discussed future plans for joint meetings with the Data Council and DSAI Institute, emphasizing collaboration and cross-disciplinary engagement in AI-related initiatives.
The meeting discussed the potential development of a service to assist researchers in utilizing large language models effectively without needing to learn complex APIs or techniques. This service aims to simplify the process and abstract the technical difficulties, allowing researchers to focus on their projects. There was interest in leveraging funding and resources from various sources, including the university, to support this initiative. Specific examples were discussed, such as classifying research studies and health plans using language models, as well as the potential application of AI in systematic reviews and meta-analyses. Overall, there was a consensus on the value of providing such a service, and plans were made to explore funding options and further develop the proposal.