Case:
ChatGPT answered my question completely wrong in a fully convincing manner. When i repeatedly told it that the answer were untrue, each response would include an excuse and a new (different but wrong) answer, stating the new (incorrect) answers as facts.
Upon addressing the fact that it keeps giving me inconsistent and incorrect answers while formulating them as facts, ChatGPT gave me an apology. It explained to me that it couldn’t actually know the answer to my question, and that it was guessing. While knowing it’s restrictions/ limits, it was formulating incorrect answers as facts, without accounting for it’s limitations.
My principle:
ChatGPT answers in a very convincing manner, stating answers as facts - without accounting for its limitations on retrieving the given answers, or on the limitations on whether it is able to know the answer.
Feedback:
- ChatGPT should be more transparent and/or honest on the probability of it’s answer.
- ChatGPT should be more transparent/honest about (it’s limitations in) retrieving certain information.
The consequences of NOT being transparent about this, and continuously stating misinformation as facts without accounting for the probability and limitations of it’s answer, is a VERY dangerous, and irresponsible practice, which should and will be taken to court.
I hope this feedback is clear.
Thank you in advance,
Daniel Mota Faria