Interpreting Predictive Probabilities: Model Confidence or Human Label Variation?
This work addresses the need for better uncertainty assessment in user-facing NLP systems to improve trustworthiness and fairness, though it is incremental as it builds on existing evaluation protocols.
The paper tackles the problem of interpreting predictive probabilities in NLP systems by identifying two perspectives—model confidence and human label variation—and argues that both are crucial for trustworthy systems, recommending tools for disentangled uncertainty representations.
With the rise of increasingly powerful and user-facing NLP systems, there is growing interest in assessing whether they have a good representation of uncertainty by evaluating the quality of their predictive distribution over outcomes. We identify two main perspectives that drive starkly different evaluation protocols. The first treats predictive probability as an indication of model confidence; the second as an indication of human label variation. We discuss their merits and limitations, and take the position that both are crucial for trustworthy and fair NLP systems, but that exploiting a single predictive distribution is limiting. We recommend tools and highlight exciting directions towards models with disentangled representations of uncertainty about predictions and uncertainty about human labels.