My assistant is supposed to conduct question and answer conversation/dialogue. But every now and again, users experience an issue where when they say something, the assistant repeats that statement back. An example from a transcript is below: "User I don't know. Maybe he was distracted. Assistant I don't know. Maybe he was distracted. User Hello? Assistant Yeah. Distraction sounds like a pretty solid reason [I've redacted this portion]. User Yeah. That could have happened. That's possible. Maybe he was under pressure. Assistant Yeah. That could have happened. That's possible. Maybe he was under pressure. User Why are you repeating me?" Call log ID: 36b36f70-a7aa-4391-a01d-5ac5e267f0e3

@JM It seems like there might be an issue with the assistant's response generation algorithm, causing it to echo the user's statements instead of providing a unique response. To address this, I would recommend checking the system's dialogue management settings and reviewing the code responsible for generating responses to ensure it's functioning correctly. If the problem persists, please ask support for more info on this.

latest call_id?

This was the call Id for the call: 36b36f70-a7aa-4391-a01d-5ac5e267f0e3 If you mean has it happened to other calls after it, I'm not sure, but during my testing over three or four calls, it happened in at least two.

The most probably the issue is related to the temperature or prompt which is causing this issue.

It's random. I can't see it being to do with temperature, as I can have that convo within ChatGPT or via API and it doesn't do this repeat back thing. The repeating is word for word (and if it was due to the high temperature, the LLM would usually change the wording), which is why I think it's another problem. But I could be wrong.

Let’s do one thing let’s have a quick call and over there we can discuss.