Technology
Welcome to the Technology Community of Practice. If you have an interest in software, hardware,... View more
AI Policies
-
AI Policies
I have seen a few examples of student unions adapting or creating their own AI Policy. To my mind, these tend to be too restrictive — setting limits or requiring a review / authorization process before any kind of AI can be used. I like to think that our existing Acceptable Use Policy covers off the worst of the potential “gotacha” concerns, and that providing information to staff is the preferred way forward.
For example, here’s my most recent post to our staff internal newsletter…
As we saw in last week’s newsletter, it is easy to upload any document for processing by CoPilot, or any other AI tool. This is partly by design, as these companies actually use the uploaded files, and your interactions with the chatbot, to train their systems. While you should still be aware of that usage, if your work does NOT involve any confidential, personal, or proprietary information, you can freely use any of the available generative AI models.However, if you do need to work with confidential, personal, or proprietary files, you should be using Microsoft’s Copilot while logged in using your @uwo.ca credentials. When logged in with your Western credentials, any data you input into Copilot receives the same protections as Western has negotiated for all the other Microsoft products. This means those uploads will not be used for general training and are not used beyond that particular session, and so would never accidentally be spit out as content in some other user’s query.As an example, if you wanted to upload a job description and a specific employee’s previous PA to help you generate a development plan you must either use Copilot, or go to the trouble of creating a separate anonymous file to upload. If you want the AI to help you analyse your budget, using Copilot while logged in means you don’t have to worry about those numbers appearing as a sample spreadsheet on someone else’s screen.Remember that AI chatbots are tools, and sharing data with them requires careful risk management on a case-by-case basis. No blanket policy can declare a particular product completely “safe,” so always consider what you are uploading.What do you think? Do you have or plan to create an AI Policy? What are your concerns either way?
Thanks,
Geoff——————————
Geoff Pimlatt
Senior Manager, Information System
University Students’ Council
University of Western Ontario
London ON
(519) 661-3574
——————————
Sorry, there were no replies found.
Log in to reply.