#MeToo Alexa: How Conversational Systems Respond to Sexual Harassment
Open accessThe DOI matches its Crossref record; 6 of the 6 references it cites were found in a registry or an index.
Abstract
Conversational AI systems are rapidly developing from purely transactional systems to social chatbots, which can respond to a wide variety of user requests.In this article, we establish how current state-of-the-art conversational systems react to inappropriate requests, such as bullying and sexual harassment on the part of the user, by collecting and analysing the novel #MeToo corpus.Our results show that commercial systems mainly avoid answering, while rule-based chatbots show a variety of behaviours and often deflect.Data-driven systems, on the other hand, are often noncoherent, but also run the risk of being interpreted as flirtatious and sometimes react with counter-aggression.This includes our own system, trained on "clean" data, which suggests that inappropriate system behaviour is not caused by data bias.Recently, widespread sexual harassment allegations following the #MeToo 3 campaign have propelled the issue of what constitutes harassment and how to respond to it to the media's attention.Given that most virtual assistants have femalesounding names and voices, it begs the question of how often these systems are harassed and how they respond to harassment (Silvervarg et al., 2012).
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Semantics derived automatically from language corpora contain human-like biases
Caliskan, A., Bryson, J. J., & Narayanan, A. (2017) · Science · doi:10.1126/science.aal4230 · Cited by 2,887
Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
Bolukbasi, T., Chang, K.-W., Zou, J., Saligrama, V., & Kalai, A. (2016) · arXiv · doi:10.48550/arxiv.1607.06520 · Cited by 1,615 · cited 2×
“…to their subjective nature and goal of mimicking human behaviour, data-driven dialogue models are susceptible to implicitly encode underlying biases in human dialogue, similar to related studies on biased lexical semantics derived from large corpora (Caliskan et al., 2017; Bolukbasi et al., 2016).”
BackgroundVinyals, O., & Le, Q. (2015) · arXiv · doi:10.48550/arxiv.1506.05869 · Cited by 1,400
Tweetment Effects on the Tweeted: Experimentally Reducing Racist Harassment
Munger, K. (2017) · Political Behavior · doi:10.1007/s11109-016-9373-5 · Cited by 395
Towards an Automatic Turing Test: Learning to Evaluate Dialogue Responses
Lowe, R., Noseworthy, M., Serban, I. V., Angelard-Gontier, N., Bengio, Y., & Pineau, J. (2017) · Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p17-1103 · Cited by 341 · cited 1×
“Current conversational AI systems are evaluated using customer satisfaction ratings, e.g. (Guo et al., 2017; Lowe et al., 2017).”
BackgroundEscaping from Children's Abuse of Social Robots
Brščić, D., Kidokoro, H., Suehiro, Y., & Kanda, T. (2015) · Proceedings of the Tenth Annual ACM/IEEE International Conference on Human-Robot Interaction · doi:10.1145/2696454.2696468 · Cited by 204 · cited 2×
“, 2015), or seeking the proximity of an authority figure (Brscić et al., 2015).”
Background
Cited by
The design and implementation of Xiaoice, an empathetic social chatbot
Zhou, L., Gao, J., Li, D., & Shum, H.-Y. (2020) · Computational Linguistics · doi:10.1162/coli_a_00368 · Cited by 777
The Design and Implementation of XiaoIce, an Empathetic Social Chatbot
Zhou, L., Gao, J., Li, D., & Shum, H.-Y. (2018) · arXiv · doi:10.48550/arxiv.1812.08989 · Cited by 777
Chaves, A. P., & Gerosa, M. A. (2021) · International Journal of Human–Computer Interaction · doi:10.1080/10447318.2020.1841438 · Cited by 572
“However, a similar number was found in a study with general purpose chatbots [Curry and Rieser 2018].”
BackgroundMethodResultInfluentialWhat Does it Mean for a Language Model to Preserve Privacy?
Brown, H., Lee, K., Mireshghallah, F., Shokri, R., & Tramèr, F. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3534642 · Cited by 165
“For example, work has been done to assess how appropriately chatbots respond to delicate situations, such as responding to sexual harassment [14], discussions of suicidal intent or other mental health issues [72, 66], and mentions of violence or physical danger [49, 66].”
BackgroundEmpathy Is All You Need: How a Conversational Agent Should Respond to Verbal Abuse
Chin, H., Molefi, L. W., & Yi, M. Y. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376461 · Cited by 96
Mirages. On Anthropomorphism in Dialogue Systems
Abercrombie, G., Curry, A., Dinkar, T., Rieser, V., & Talat, Z. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.290 · Cited by 83
“In such circumstances systems have even been shown to respond subserviently to their abusers, potentially further encouraging the behaviour (Cer-cas Curry and Rieser, 2018).”
BackgroundSelf-citationThe Illusion of Empathy? Notes on Displays of Emotion in Human-Computer Interaction
Cuadra, A., Wang, M., Stein, L. A., Jung, M. F., Dell, N., Estrin, D., & Landay, J. A. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642336 · Cited by 80
“Unfortunately, in line with prior work [14], they found that the CAs responded inconsistently and incompletely.”
Sexual Interaction in Digital Contexts and Its Implications for Sexual Health: A Conceptual Analysis
Döring, N., Krämer, N., Mikhailova, V., Brand, M., Krüger, T. H. C., & Vowe, G. (2021) · Frontiers in Psychology · doi:10.3389/fpsyg.2021.769732 · Cited by 62
“Issues of consent arise when sexbots appear childlike or when users engage in aggressive and abusive interactions with virtual assistants (Laorden et al., 2015; Curry and Rieser, 2018).”
BackgroundLlama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
Inan, H., Upasani, K., Chi, J., Rungta, R., Iyer, K., Mao, Y., Tontchev, M., Hu, Q., Fuller, B., Testuggine, D., et al. (2023) · arXiv · doi:10.48550/arxiv.2312.06674 · Cited by 56
“In addition to checking human-generated content, making LLM-based dialog systems safe requires checking model responses, as the system may generate inappropriate content (Dinan et al., 2019), or respond inappropriately to offensive content (Lee et al., 2019; Cercas Curry and Rieser, 2018).”
BackgroundTheories of ”gender” in NLP bias research
Devinney, H., Björklund, J., & Björklund, H. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3534627 · Cited by 49
“Date me date me”: AI chatbot interactions as a resource for the online construction of masculinity
Koh, J. (2023) · Discourse, Context & Media · doi:10.1016/j.dcm.2023.100681 · Cited by 15
FairPrism: Evaluating Fairness-Related Harms in Text Generation
Fleisig, E., Amstutz, A., Atalla, C., Blodgett, S. L., Daumé III, H., Olteanu, A., Sheng, E., Vann, D., & Wallach, H. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.343 · Cited by 14
“…AI system exacerbates the harms in the human input, counters the harms, or attempts to avoid the topic affects the extent to which harms are present in the interaction as a whole, even if the AI-generated output seems innocuous in isolation (Schlesinger et al., 2018; Cercas Curry and Rieser, 2018).”
BackgroundInfluentialCloud-Based Sentiment Analysis for Interactive Agents
Keijsers, M., Bartneck, C., & Kazmi, H. S. (2019) · Proceedings of the 7th International Conference on Human-Agent Interaction · doi:10.1145/3349537.3351883 · Cited by 9
“come more ubiquitous - already, people use their Alexa or Siri for a wide range of conversation topics which stretches far beyond mere inquiries about the weather or news headlines [11, 14].”
Background(2019) · AICS
Metadata
- Authors
- Cercas Curry, A., & Rieser, V.
- Venue
- Proceedings of the Second ACL Workshop on Ethics in Natural Language Processing
- Published
- 2018
- Pages
- 7-14
- OpenAlex
- W2806344213
- Cited by
- 86
- References
- 6
- FWCI
- 19.91
- Type
- conference paper
- Language
- en
- Open access
- gold
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.