Skip to main content
. 2020 Mar 31;10:5732. doi: 10.1038/s41598-020-61994-0

Figure 3.

Figure 3

Conversational efficiency of AI agents. The x-axis represents the number of dialogue turns elapsed. The y-axis represents various performance metrics. Baseline refers to the performance of MCI classifier using all the responses generated from the original dataset. By contrast, RL refers to the performance of MCI classifier using responses generated by the user simulator, in response to the agent-generated questions at test time.