From Lived Experience to Insight: Unpacking the Psychological Risks of Using AI Conversational Agents

Recent gain in popularity of AI conversational agents has led to their increased use for improving productivity and supporting well-being. While previous research has aimed to understand the risks associated with interactions with AI conversational agents, these studies often fall short in capturing the lived experiences. Additionally, psychological risks have often been presented as a sub-category within broader AI-related risks in past taxonomy works, leading to under-representation of the impact of psychological risks of AI use. To address these challenges, our work presents a novel risk taxonomy focusing on psychological risks of using AI gathered through lived experience of individuals. We employed a mixed-method approach, involving a comprehensive survey with 283 individuals with lived mental health experience and workshops involving lived experience experts to develop a psychological risk taxonomy. Our taxonomy features 19 AI behaviors, 21 negative psychological impacts, and 15 contexts related to individuals. Additionally, we propose a novel multi-path vignette based framework for understanding the complex interplay between AI behaviors, psychological impacts, and individual user contexts. Finally, based on the feedback obtained from the workshop sessions, we present design recommendations for developing safer and more robust AI agents. Our work offers an in-depth understanding of the psychological risks associated with AI conversational agents and provides actionable recommendations for policymakers, researchers, and developers.

Lived Experience Not Found: LLMs Struggle to Align with Experts on Addressing Adverse Drug Reactions from Psychiatric Medication Use

Adverse Drug Reactions (ADRs) from psychiatric medications are the leading cause of hospitalizations among mental health patients. With healthcare systems and online communities facing limitations in resolving ADR-related issues, Large Language Models (LLMs) have the potential to fill this gap. Despite the increasing capabilities of LLMs, past research has not explored their capabilities in detecting ADRs related to psychiatric medications or in providing effective harm reduction strategies. To address this, we introduce the Psych-ADR benchmark and the Adverse Drug Reaction Response Assessment (ADRA) framework to systematically evaluate LLM performance in detecting ADR expressions and delivering expert-aligned mitigation strategies. Our analyses show that LLMs struggle with understanding the nuances of ADRs and differentiating between types of ADRs. While LLMs align with experts in terms of expressed emotions and tone of the text, their responses are more complex, harder to read, and only 70.86% aligned with expert strategies. Furthermore, they provide less actionable advice by a margin of 12.32% on average. Our work provides a comprehensive benchmark and evaluation framework for assessing LLMs in strategy-driven tasks within high-risk domains.

MedHalu: Hallucinations in Responses to Healthcare Queries by Large Language Models

The remarkable capabilities of large language models (LLMs) in language understanding and generation have not rendered them immune to hallucinations. LLMs can still generate plausible-sounding but factually incorrect or fabricated information. As LLM-empowered chatbots become popular, laypeople may frequently ask health-related queries and risk falling victim to these LLM hallucinations, resulting in various societal and healthcare implications. In this work, we conduct a pioneering study of hallucinations in LLM-generated responses to real-world healthcare queries from patients. We propose MedHalu, a carefully crafted first-of-its-kind medical hallucination dataset with a diverse range of health-related topics and the corresponding hallucinated responses from LLMs with labeled hallucination types and hallucinated text spans. We also introduce MedHaluDetect framework to evaluate capabilities of various LLMs in detecting hallucinations. We also employ three groups of evaluators -- medical experts, LLMs, and laypeople -- to study who are more vulnerable to these medical hallucinations. We find that LLMs are much worse than the experts. They also perform no better than laypeople and even worse in few cases in detecting hallucinations. To fill this gap, we propose expert-in-the-loop approach to improve hallucination detection through LLMs by infusing expert reasoning. We observe significant performance gains for all the LLMs with an average macro-F1 improvement of 6.3 percentage points for GPT-4.

Better to Ask in English: Cross-Lingual Evaluation of Large Language Models for Healthcare Queries

Large language models (LLMs) are transforming the ways the general public accesses and consumes information. Their influence is particularly pronounced in pivotal sectors like healthcare, where lay individuals are increasingly appropriating LLMs as conversational agents for everyday queries. While LLMs demonstrate impressive language understanding and generation proficiencies, concerns regarding their safety remain paramount in these high-stake domains. Moreover, the development of LLMs is disproportionately focused on English. It remains unclear how these LLMs perform in the context of non-English languages, a gap that is critical for ensuring equity in the real-world use of these systems.This paper provides a framework to investigate the effectiveness of LLMs as multi-lingual dialogue systems for healthcare queries. Our empirically-derived framework XlingEval focuses on three fundamental criteria for evaluating LLM responses to naturalistic human-authored health-related questions: correctness, consistency, and verifiability. Through extensive experiments on four major global languages, including English, Spanish, Chinese, and Hindi, spanning three expert-annotated large health Q&A datasets, and through an amalgamation of algorithmic and human-evaluation strategies, we found a pronounced disparity in LLM responses across these languages, indicating a need for enhanced cross-lingual capabilities. We further propose XlingHealth, a cross-lingual benchmark for examining the multilingual capabilities of LLMs in the healthcare context. Our findings underscore the pressing need to bolster the cross-lingual capacities of these models, and to provide an equitable information ecosystem accessible to all.

Understanding the Humans Behind Online Misinformation: An Observational Study Through the Lens of the COVID-19 Pandemic

The proliferation of online misinformation has emerged as one of the biggest threats to society. Considerable efforts have focused on building misinformation detection models, still the perils of misinformation remain abound. Mitigating online misinformation and its ramifications requires a holistic approach that encompasses not only an understanding of its intricate landscape in relation to the complex issue and topic-rich information ecosystem online, but also the psychological drivers of individuals behind it. Adopting a time series analytic technique and robust causal inference-based design, we conduct a large-scale observational study analyzing over 32 million COVID-19 tweets and 16 million historical timeline tweets. We focus on understanding the behavior and psychology of users disseminating misinformation during COVID-19 and its relationship with the historical inclinations towards sharing misinformation on Non-COVID topics before the pandemic. Our analysis underscores the intricacies inherent to cross-topic misinformation, and highlights that users' historical inclination toward sharing misinformation is positively associated with their present behavior pertaining to misinformation sharing on emergent topics and beyond. This work may serve as a valuable foundation for designing user-centric inoculation strategies and ecologically-grounded agile interventions for effectively tackling online misinformation.

Towards Massively Multi-domain Multilingual Readability Assessment

We present ReadMe++, a massively multi-domain multilingual dataset for automatic readability assessment. Prior work on readability assessment has been mostly restricted to the English language and one or two text domains. Additionally, the readability levels of sentences used in many previous datasets are assumed on the document-level other than sentence-level, which raises doubt about the quality of previous evaluations. We address those gaps in the literature by providing an annotated dataset of 6,330 sentences in Arabic, English, and Hindi collected from 64 different domains of text. Unlike previous datasets, ReadMe++ offers more domain and language diversity and is manually annotated at a sentence level using the Common European Framework of Reference for Languages (CEFR) and through a Rank-and-Rate annotation framework that reduces subjectivity in annotation. Our experiments demonstrate that models fine-tuned using ReadMe++ achieve strong cross-lingual transfer capabilities and generalization to unseen domains. ReadMe++ will be made publicly available to the research community.

"Subverting the Jewtocracy": Online Antisemitism Detection Using Multimodal Deep Learning

The exponential rise of online social media has enabled the creation, distribution, and consumption of information at an unprecedented rate. However, it has also led to the burgeoning of various forms of online abuse. Increasing cases of online antisemitism have become one of the major concerns because of its socio-political consequences. Unlike other major forms of online abuse like racism, sexism, etc., online antisemitism has not been studied much from a machine learning perspective. To the best of our knowledge, we present the first work in the direction of automated multimodal detection of online antisemitism. The task poses multiple challenges that include extracting signals across multiple modalities, contextual references, and handling multiple aspects of antisemitism. Unfortunately, there does not exist any publicly available benchmark corpus for this critical task. Hence, we collect and label two datasets with 3,102 and 3,509 social media posts from Twitter and Gab respectively. Further, we present a multimodal deep learning system that detects the presence of antisemitic content and its specific antisemitism category using text and images from posts. We perform an extensive set of experiments on the two datasets to evaluate the efficacy of the proposed system. Finally, we also present a qualitative analysis of our study.

AbuseAnalyzer: Abuse Detection, Severity and Target Prediction for Gab Posts

While extensive popularity of online social media platforms has made information dissemination faster, it has also resulted in widespread online abuse of different types like hate speech, offensive language, sexist and racist opinions, etc. Detection and curtailment of such abusive content is critical for avoiding its psychological impact on victim communities, and thereby preventing hate crimes. Previous works have focused on classifying user posts into various forms of abusive behavior. But there has hardly been any focus on estimating the severity of abuse and the target. In this paper, we present a first of the kind dataset with 7601 posts from Gab which looks at online abuse from the perspective of presence of abuse, severity and target of abusive behavior. We also propose a system to address these tasks, obtaining an accuracy of ~80% for abuse presence, ~82% for abuse target prediction, and ~65% for abuse severity prediction.

Exact Passive-Aggressive Algorithms for Learning to Rank Using Interval Labels

In this paper, we propose exact passive-aggressive (PA) online algorithms for learning to rank. The proposed algorithms can be used even when we have interval labels instead of actual labels for examples. The proposed algorithms solve a convex optimization problem at every trial. We find exact solution to those optimization problems to determine the updated parameters. We propose support class algorithm (SCA) which finds the active constraints using the KKT conditions of the optimization problems. These active constrains form support set which determines the set of thresholds that need to be updated. We derive update rules for PA, PA-I and PA-II. We show that the proposed algorithms maintain the ordering of the thresholds after every trial. We provide the mistake bounds of the proposed algorithms in both ideal and general settings. We also show experimentally that the proposed algorithms successfully learn accurate classifiers using interval labels as well as exact labels. Proposed algorithms also do well compared to other approaches.