# Probing Neural Dialog Models for Conversational Understanding

arXiv: https://arxiv.org/abs/2006.08331 · HTML: https://mlsanity.com/paper/2006.08331/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Jun 2020.

## Built on by (most cited first, 10 of 10)

- [Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks](https://mlsanity.com/paper/2207.13243/) (2022)
- [Investigating Robustness of Dialog Models to Popular Figurative Language Constructs](https://mlsanity.com/paper/2110.00687/) (2021)
- [Operationalising Representation in Natural Language Processing](https://mlsanity.com/paper/2306.08193/) (2023)
- [Probing via Prompting](https://mlsanity.com/paper/2207.01736/) (2022)
- [Do Charge Prediction Models Learn Legal Theory?](https://mlsanity.com/paper/2210.17108/) (2022)
- [Can Visual Dialogue Models Do Scorekeeping? Exploring How Dialogue Representations Incrementally Encode Shared Knowledge](https://mlsanity.com/paper/2204.06970/) (2022)
- [Neural Models and Language Model Prompting for the Multidimensional Evaluation of Open-Ended Conversations](https://mlsanity.com/paper/2509.00841/) (2025)
- [Visualizing the Relationship Between Encoded Linguistic Information and Task Performance](https://mlsanity.com/paper/2203.15860/) (2022)
- [A survey of neural models for the automatic analysis of conversation: Towards a better integration of the social sciences](https://mlsanity.com/paper/2203.16891/) (2022)
- [An Empirical Study on Context Length for Open-Domain Dialog Generation](https://mlsanity.com/paper/2409.00315/) (2024)

## Abstract (arXiv)

The predominant approach to open-domain dialog generation relies on end-to-end training of neural models on chat datasets. However, this approach provides little insight as to what these models learn (or do not learn) about engaging in dialog. In this study, we analyze the internal representations learned by neural open-domain dialog systems and evaluate the quality of these representations for learning basic conversational skills. Our results suggest that standard open-domain dialog systems struggle with answering questions, inferring contradiction, and determining the topic of conversation, among other tasks. We also find that the dyadic, turn-taking nature of dialog is not fully leveraged by these models. By exploring these limitations, we highlight the need for additional research into architectures and training methods that can better capture high-level information about dialog.