arXiv ScienceSearch

arXiv · 2402.07920

Exploring patient trust in clinical advice from AI-driven LLMs like ChatGPT for self-diagnosis

Abstract

Trustworthy clinical advice is crucial but can be burdensome to obtain. Limited access and financial costs may lead people to self-diagnose. However, self-diagnosis requires considerable learning and can create risks when people pursue treatment without professional guidance. Large language models (LLMs) such as GPT-4 may offer a convenient yet risky alternative because they can produce inaccurate but convincing information. We therefore ask whether patients can trust clinical advice from AI-driven LLMs. We examined this question through a think-aloud observation in which a patient used GPT-4 for self-diagnosis while a doctor assessed its responses using professional expertise. We then conducted a semi-structured interview with the patient about their trust in the system. Our results show that patients may struggle to identify errors because they lack professional medical knowledge, even when GPT-4 provides advice that a doctor can recognize as false. Patients may develop some trust because GPT-4 explains its responses and acknowledges its limitations, but this trust remains uncertain because its advice can be unreliable. The doctor also reported that checking GPT-4's responses required more effort than making a diagnosis without it. Patients tend to trust doctors because educated and authorized professionals can provide effective guidance. This trust also develops through social connection, certification, institutional accountability, and professional rules. Doctors can adapt their questions, observe patients, and use different methods when patients cannot clearly describe their symptoms. An LLM, however, depends primarily on the information provided in a prompt and may overlook details that a doctor could identify during a clinical consultation. These findings raise questions about competence, responsibility, autonomy, and safety when LLMs are used for clinical advice.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Delong Du, Richard Paluch, Gunnar Stevens, Claudia Müller. 2026-09-07. Exploring patient trust in clinical advice from AI-driven LLMs like ChatGPT for self-diagnosis. https://arxiv.org/abs/2402.07920

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related papers

Are explainable AI (XAI) evaluation strategies aligned? Comparing subjective, objective, and mathematical evaluation measures using saliency maps

The evaluation of explainable AI (XAI) approaches often relies on three families of methods: subjective measures (e.g., questionnaires on trust or satisfaction), objective measures (e.g., task performance metrics), and mathematical metrics (e.g., for faithfulness). Yet, it remains unclear how these families align or diverge in practice. In a{preregistered} between-subjects study (N=166), we use three established saliency map techniques (LIME, Grad-CAM, Guided Backpropagation) as a testbed to examine this issue. We find that each family of methods leads to different conclusions: participants reported no differences in trust or satisfaction, Grad-CAM improved user performance, while mathematical metrics favored Guided Backpropagation. At the same time, mathematical metrics were only partially related to user performance, and these relationships were sometimes counterintuitive. Our findings highlight the methodological importance of comparing subjective, objective, and mathematical approaches when evaluating XAI, illustrating both tensions and aspects that are aligned. We discuss implications for XAI evaluation frameworks.

cs.HC

Mind Your Ps and Qs: Positive Moderation Practice in the Positive Queue

Online communities rely on volunteer moderators to maintain order. Despite their key role, moderators are given a toolbox of punishments and far less support for encouraging contributions they want to see more of. We introduce the Positive Queue as a positive counterpart to Reddit's modqueue: a dedicated space for moderators to discover contributions and behaviors they want to encourage and positively reinforce. With five moderators, four with 6-14 years of experience, we use the Positive Queue to examine how moderators operationalize positive reinforcement. Moderators combined predicted community reception, observed engagement, and their own judgment; used prediction-engagement mismatches to identify overlooked content; and repurposed positive features for punitive and retrospective work. These findings surface tensions around labor, attribution, and community fit. We contribute the Positive Queue as a working system and conceptualize positive moderation as recognition infrastructure that shapes what moderators notice, whose judgment becomes visible, and how recognition reaches contributors.

cs.HC

PILOT: Control Surfaces for Authoring Social Media Feeds

Personalized social media feeds infer preferences from behavior, leaving people little direct control over what they see. Existing controls range from post-level reactions to rules and natural language, but little is known about how people use them together or how added expressiveness changes effort. We built PILOT, a Bluesky feed-authoring system that turns in-feed actions into explicit preferences, deterministic ranking, and inspectable outcomes. A study with seven participants informed an expanded implementation, which we organized into three nested control surfaces that ten participants compared within subjects. Participants assigned controls to distinct jobs: broad controls set direction, in-post actions refined results, and filters removed content. Richer surfaces did not necessarily feel more effortful, and participants' experiences depended more on whether they could verify and repair outcomes. Our findings show that usable feed control requires not simply more controls, but orchestration across mechanisms that support expression, inspection, repair, and episodic use.

cs.HC