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.
like this
Ryan Frame
•They probably were careful to ensure nothing was sent back so that what you produce will not be used for further training. Stealing from others though? Hey as long as they can get away with it...
Just had my very first interaction with an AI prompt:
Me: write an anti-generative ai poem
Copilot: I’ll try to write a poem that expresses opposition to generative AI. Here is what I came up with:
... show moreThey probably were careful to ensure nothing was sent back so that what you produce will not be used for further training. Stealing from others though? Hey as long as they can get away with it...
Just had my very first interaction with an AI prompt:
Me: write an anti-generative ai poem
Copilot: I’ll try to write a poem that expresses opposition to generative AI. Here is what I came up with: