AI Features FAQ

This documents answers frequently asked questions about AI features in online services that UIS provides, such as Zoom, Google and Box.

Contents

 

Q: When will AI features be available?

A: UIS will make AI features in Google, Box and Zoom available May 21st, 2024.  New features, as they become available, will be made available as UIS is able to review them.

 

Q: Who can utilize AI features?

A: Anyone with accounts on Pacific University's Box, Google and regular Zoom (not Protected Data Zoom) can access these features.  Generally speaking, that means faculty, staff and students can access these features, as well as sponsored accounts where that access has been specifically requested.

 

Q: How do I know that AI features are secure and private?

A: Before turning on AI features in any of our core services (the major services maintained by UIS for the use and benefit of the university) UIS does an evaluation of the privacy and security of that data.  See our AI Evaluation Procedures.  AI features that UIS has not evaluated may not be secure or protect the privacy of the information they process.

 

Q: Are AI features available in Protected Data Zoom?

A: No, because Protected Data Zoom is used for discussions that include Patient Health Information, this puts is under the scope of HIPAA's security and privacy compliance standards, and so we are not at this time making this feature available there.

 

Q: Can I use sensitive or protected data with AI features?

A: If you can store a piece of data in a Pacific University Box, Zoom or Google, then you can use the AI features on that application with that data.  For example, Box is authorized for use with FERPA protected data and with HIPAA protected data (in a HIPAA-configured folder) and so AI can be used with these.  Google and regular Zoom (not Protected Data Zoom) are authorized for FERPA protected data but not HIPAA protected data, so one should not use the AI features in these apps with HIPAA protected data.  Use of AI products and services that UIS has not evaluated is strictly prohibited.

 

Q: How do I find and utilize AI features?

In Box, click the colorful-atom icon at the top of the screen while viewing a document (PDF, Word, text, etc.) in Box preview on the Box web interface.  Or, on a Box Note, use your mouse to select text and then click the same icon on the small bar that appears.

In Zoom, log into your Zoom account via the web then go to your Settings page and go to the AI Companion tab and turn on features you want to use.  During a meeting, you can click the Start Summary icon (page with diamond shapes) at the bottom of the screen.  If you've turned on "Smart Recording with AI Companion" then starting cloud recording will automatically engage the AI.

In Google, you can go to gemini.google.com when logged in with your Pacific account.

 

Q: Am I limited in how much I can utilize AI features?

For Zoom and Google you are not limited in how much you can use AI features.

For Box, each user has 20 queries per month.  After a user has used their queries for the month, they start using queries from a shared pool that all Box users have access to.  There is currently no way to see, in Box, how many queries one has used for the month.  UIS asks every user to be conscientious in their use of queries so as not to deprive others of the ability to use this resource.

 

Q: Is there anything I'm not allowed to use AI features for?

As discussed above, if any information should not be stored in these services (e.g. Patient Health Information should not be stored in Google or discussed over regular Zoom) then it is a violation of policy to utilize the AI features in these services with that data.  Use of protected or sensitive data in any AI service UIS has not evaluated is strictly prohibited.

Box's Acceptable Use Policy for AI, which all Box users are bound to by using the AI service, prohibits the use of Box AI to make decisions about:

  • Employment
  • Healthcare
  • Finance
  • Legal Matters
  • Education
  • Any decisions made based on people's legally defined protected classes or characteristics (e.g. race, ethnicity, religion)

The specific language in the policy is that Box AI "should not be used to make [decisions]" concerning the categories above.  They also mention not "relying upon" the AI in these matters.  These prohibitions fit well with UIS' recommendation on the use of all AI tools: since AI can be both inaccurate and biased (see below) it can be used to generate ideas and draft text, but its recommendations and text should be thoroughly checked by a qualified individual before being acted upon in any way.

Furthermore, Box prohibits using Box AI to assist in any unsafe or illegal behaviors.  They also prohibit any attempt to override the safety filters meant to stop Box from offering information that contravenes this policy.

For students: your instructors may prohibit the use of AI in your schoolwork.  Check your syllabus or ask your instructor before using any of these tools.

The copyright status of the text or images generated by these tools is still unclear.  Do not include text or images in any product that you wish to copyright, publish or sell.

Utilizing text or images created by an AI without crediting the source of those images may count as plagiarism and may violate various university policies concerning honesty and academic integrity.

Where university policy or practice prohibits the recording of an event (such as certain meetings where university governance policies specify that they are not to be recorded), using the Zoom AI Meeting Assistant to generate a summary of the meeting may count as recording that meeting.

External entities you may be working with may have their own prohibitions about the use of AI tools.  For example, the National Institutes of Health prohibits the use of any generative AI tools for anyone engaging in their peer review process for proposals.

 

Q: Will AI features give access to my private info or communications to other Pacific University users?

A: In Google, the Gemini feature only has access to publicly available information.

In Box, when you use the Box AI your queries are only able to access Box files that you yourself have access to.  Nobody else can do queries referencing documents you control unless you have given them access to those documents.

In Zoom, when you turn on the "Meeting Summary with AI Companion" feature, you are given the following options for who the meeting summary is shared with:

  • Only me (meeting host)
  • Only myself (meeting host) and meeting invitees in our organization
  • All meeting invitees including those outside of our organization

If you choose the latter two, then a person who joins your meeting later may automatically receive a summary that may let them know about things discussed while they were not present, and which meeting attendees might not have wanted to discuss had that person not been present.

 

Q: Are the answers these AI services generate accurate?

A: No.  AI tools are known to give incorrect or sometimes completely fictional information, presented as fact, and have done so even in UIS' testing.  Any information given by AI tools should be fact-checked before being acted upon in any meaningful way.

 

Q: Are the answers these AI services give unbiased?

A: No.  AI models are trained on text and images publicly available on the Internet, and thus can reflect the biases shown by the humans who created those original text and images.  Users of AI services should always be wary that these biases may surface in subtle ways in the content that AI services generate.  

 

Q: How can I request that a new AI feature or tool be made available?

A: Use the "Request Evaluation, Security Review & Approval for a Purchase" form on this page to ask us to take a look at a product you'd like to use.  For anything that may qualify as a UIS project, the "Step One: Technology Change/Adoption/Need Query" form on this page should be used.

 

Q: How do I delete the content created by these AI features?

A: In Box, closing the document deletes your chat with the Box AI and that text will be gone unless you took positive steps to save it elsewhere.

In Zoom, to delete meeting summaries, log into the Zoom web interface and click "AI Companion" on the left.  Enter a check mark to the left of any meeting summary you want deleted and click the "Delete Selected" button.  To delete meeting summaries embedded in Zoom recordings, log into the Zoom web interface and click Recordings on the left.  Enter a check mark to the left of any meeting recording you want deleted and click the "Delete Selected" button. 

In Google, go to the Google Gemini web interface, and click "Activity" in the bottom left.  Find the entry for the prompt and it's response and click the "x" next to it to permanently delete your prompt (including any images you supplied) and Google's response.

 

See Also

AI Evaluation Procedures

Questions?

Contact Support

Details

Article ID: 150375
Created
Thu 4/4/24 6:30 PM
Modified
Fri 4/19/24 12:41 PM