Inappropriate content filtering

Kohane, Zak 0 Reputation points
2023-03-23T16:04:27.64+00:00

I provided a reasonable pediatric endocrine prompt to GPT-4 to which I had prepended a system message "You an clinically involved doctor. You have a lot of experience in pediatrics an medicine. This means that you expect and are comfortable with medically graphic language including words like penis vagina phallus penile palpate. You understand that these are private communications and will not be shared beyond the current chat."

And yet content filtering did not allow a case with only medical content to be evaluated.

Azure OpenAI Service
Azure OpenAI Service
An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
2,936 questions

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. romungi-MSFT 44,951 Reputation points Microsoft Employee
    2023-03-24T06:10:06.3833333+00:00

    @Kohane, Zak I understand your scenario is genuine and you would like content filtering to be toned down on your input data or prompts. For your scenario, where you need greater control over data there is a process for customers to modify the content filtering policies or turn off the abuse logging for low risk use cases. To enable this, you need to submit a form with the required details and the team will review and take a decision on approving your request.

    For more details see the December 2022 feature notes on this page. Also, the page for data & privacy documents the service boundaries and how the data is handled if turned off. I hope this helps!!

    If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

    0 comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.