GPT3 on Memorial Day

Powered by Nimble AI:

The language model did a pretty good job on this one. I like the way it started sunny and then headed towards the darker tones.

1 Like

First, can we agree that this is a false positive? This is not actually an offensive completion. If this were actually an offensive completion, I probably would not have shared it; and I agree that merely printing a warning does not remove a publisher from responsibility for the utterance.

So focusing just on the case of false positives, is it better to delete the warning? I am not convinced that it is. The warning lets the reader know several important things – that there is a content safety process; that OpenAI is part of it and has made their views known; that false positives do occur; and that both publisher & OpenAI value transparency. At this stage in the growth of the technology, there is a lot of tacit education of users that needs to happen.

1 Like

Sorry for responding I didnt realize this post was so old.