AI Capture Guidelines

Background:

Artificial Intelligence (AI) continues to evolve rapidly. Therefore, this guidance may be updated frequently and without advanced notice.

The use of AI assistants for classes and meetings can offer significant benefits, including automation of note-taking, scheduling, and follow-up tasks. AI assistants can transcribe discussions, highlight key points, and distribute action items, with the potential of enhancing productivity and capturing critical details. These tools can also analyze content to provide insights and suggest improvements, making meetings more effective. Through these functions, AI can alleviate some of the administrative burdens on educators, researchers, and administrators, allowing them to focus more on providing their specific insight and expertise, and less on clerical tasks. 

However, there are risks associated with the use of AI to capture meetings and classes that range from privacy and ethics to equity, access, critical thinking, and potential for inaccuracy and miscommunication. A few key considerations include:

  • Privacy and security concerns are paramount in order to prevent unintentional exposure of sensitive information discussed in meetings and/or the use of data to train AI systems without appropriate authorizations or other required controls.
  • AI models that are trained on data sets that are not inclusive can return results that are inaccurate or exacerbate inequality and discrimination.
  • Privacy concerns must also be balanced with disability accommodations.
  • Over-reliance on AI might reduce critical thinking, and more importantly, reduce human oversight.
  • Inaccuracies in transcription or data analysis could lead to misunderstandings, poor decision-making, or legal liability.

Ensuring the ethical use of AI and maintaining a balance between automation and human input is crucial for maximizing the benefits while mitigating the risks.  

Except in contexts where specific regulations are in effect, these guidelines are not prescriptive and are intended to assist UNM faculty, staff, and students in balancing the benefits of these technologies while reducing the risks associated with their use.  Questions should be submitted via Help.UNM to the UNM Information Security and Privacy Office (ISPO). 

Definitions and Categories of Artificial Intelligence (AI) Capturing of Meetings: 

 Virtual or in-person meetings where content of audio or video conferences or calls falls within the following 4 categories:

  1. Academic/Class meeting: Content that is of a classroom, educational, testing, or pedagogical nature
  2. Administrative meeting: Content that is of a university business nature
  3. Clinical/ Patient meeting: Content that contains, involves, or relates to Protected Health Information (PHI) or that is of a patient or patient care nature
  4. Research meeting: Content that that contains, involves, or relates to a funded or unfunded sponsored work nature
  5. Artificial Intelligence (AI) and AI Systems: Software system that can make predictions, recommendations, recognitions, follow prompts, or make decisions influencing real or virtual environments. These include, but are not limited to Zoom AI Companion, Otter.ai, and Microsoft CoPilot
  6. Capture/ Capturing: Reduction or extraction of contents of a meeting or class to an alternative tangible form by automated and/ or technical means. Capturing includes:
    • Recording (i.e., standard video recording without AI features), 
    • Smart recordings (e.g., Zoom Smart Recording),  
    • Using AI to summarize (e.g., Zoom Meeting Summary), and
    • Transcribing  

Special thanks to the Chief Privacy Office of the University of California, San Diego (UCSD) for generously sharing the content for the vast majority of this guideline.