I understand why people are somewhat frustrated by the "safety bumpers" on this. But I would say that I am actually really impressed by the quality of those safety controls. This is an AI that seems to know what it can and can not give a decent response to. I don't know if that's hard coded or trained in, but it's really impressive when you compare it to the hallucinations that typically show up in GPT3.
For example, ChatGPT refused to give me an opinion on new technologies (it says it doesn't have enough recent information) or political figures (it says it doesn't have enough personal experience). In contrast, it's happy to tell me what I should buy for a white-elephant at my office.
This is way, way closer to being useful than a model which just answers every question, regardless of whether it knows the answer.
For example, ChatGPT refused to give me an opinion on new technologies (it says it doesn't have enough recent information) or political figures (it says it doesn't have enough personal experience). In contrast, it's happy to tell me what I should buy for a white-elephant at my office.
This is way, way closer to being useful than a model which just answers every question, regardless of whether it knows the answer.