Conversation

My company recently internally promoted using Microsoft Copilot for Web including to "Generate creative content (such as poems, stories, songs, and more).". I replied to the Slack message some snarky questions about author consent to have their work used by the LLM tool, after which I was directed to an internal feedback form about LLM usage.

I repeated my question "Do we have the details on the dataset Copilot was trained on and if the relevant authors were asked for their work to be used, especially for the “creative content generation” part?" and I received a personal response from the Senior Vice President of End User Technology & Experience that "No data is tracked, sent to LLM models and microsoft doesnt (sic) know what we ask and return".

I was initially only venting, but the escalation culminating in a non sequitur has made me even more frustrated.

#LLM #AI