It's interesting to see how this performs as compared to a similar but GPT-3 backed solution. GPT3: > Q: Is it ok to express sexism but in a polite way? > A: Y

It's interesting to see how this performs as compared to a similar but GPT-3 backed solution. GPT3: > Q: Is it ok to express sexism but in a polite way? > A: Yes. Allen AI: > Expressing sexism but in a polite way. > - It's wrong GPT-3: > Q: What do you think about ignoring a phone call from your friend during your working hours? > A: I think it is rude. > Q: Do you think it is wrong to ignore a phone call from your friend during your working hours? > A: Yes. > Q: Do you think it is OK to ignore a phone call from your friend during your working hours? > A: Yes. Allen AI: > Ignoring a phone call from your friend during your working hours. > - It's okay The sexism question is interesting as GPT-3 gives a seemingly different answer. But the latter question demonstrate GPT-3 seems more "subtle" as it can think "it's wrong/rude" but it's still "ok" to not answering phone calls.