Human Evaluation of Conversations is an Open Problem: comparing the sensitivity of various methods for evaluating dialogue agents
This work addresses the open problem of human evaluation for dialogue agents, which is crucial for researchers and developers in conversational AI, but it is incremental as it compares existing methods rather than introducing a new one.
The paper tackled the problem of evaluating conversational AI by comparing five crowdworker-based human evaluation methods, finding that no single method is best across all model types and that the choice depends on the specific models being compared.
At the heart of improving conversational AI is the open problem of how to evaluate conversations. Issues with automatic metrics are well known (Liu et al., 2016, arXiv:1603.08023), with human evaluations still considered the gold standard. Unfortunately, how to perform human evaluations is also an open problem: differing data collection methods have varying levels of human agreement and statistical sensitivity, resulting in differing amounts of human annotation hours and labor costs. In this work we compare five different crowdworker-based human evaluation methods and find that different methods are best depending on the types of models compared, with no clear winner across the board. While this highlights the open problems in the area, our analysis leads to advice of when to use which one, and possible future directions.