Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2360
version: 2
kind: revised
reason: Editorial pass: sharper headline; gain/problem restated as concrete outcomes.
timestamp: 2026-11-04T09:49:07.033998Z
status: published
lens: g_space
sector: health
headline: Patient trust in medical AI: Conditions under which chatbots are preferred over human providers
dek: Background Artificial intelligence (AI) tools, including medical chatbots, are increasingly integrated into health care delivery. While prior research emphasizes diagnostic accuracy and usability, less is known about the psychological and contextual factors that shape patients' trust and preferences for AI versus human providers, particularly among patients who have experienced medical stigma. Purpose This study investigates when and why patients may prefer medical chatbots over human providers, focusing on the…
gain_title: In two hypothetical-scenario experiments with 750 participants, those with past medical stigma or more comfort with medical AI reported greater trust in a chatbot, though overall people preferred human providers.
problem_title: (none)
trace_subject: (none)
gain_reading: In two hypothetical-scenario experiments with 750 participants, those with past medical stigma or more comfort with medical AI reported greater trust in a chatbot, though overall people preferred human providers.
gain_evidence: participants who reported high comfort with medical AI and those with prior experiences of stigma (e.g., related to weight, mental health, or gender identity) expressed greater trust in the chatbot | Chatbots may offer psychological safety and perceived impartiality for patients with stigmatized identities
problem_reading: (none)
problem_evidence: (none)
quick_read: Two scenario-based experiments tested how patients judge a medical chatbot against a human provider. One compared trust in each provider type, and the other examined which one people chose when wait times differed. Trust was measured through perceived accuracy and bias, alongside past stigma experiences and comfort with medical AI.

The results suggest that acceptance of AI in care is not uniform. Chatbots might help widen access and build trust among marginalized groups, particularly for low-stakes concerns. Because the work relies on hypothetical scenarios rather than real care encounters, it remains unclear whether the stated preferences would carry over to actual clinical settings.
limitation: The findings come from scenario-based experiments, and the authors frame the benefits as applying only to selected contexts, especially low-stakes concerns, rather than care in general.
tag: Evidence-backed gain
key_points: Overall, participants preferred human providers to the medical chatbot. | Higher comfort with medical AI was linked to greater trust in the chatbot. | Prior experiences of stigma, such as related to weight, mental health, or gender identity, were also linked to greater chatbot trust. | Moderation analysis found significant interactions between provider type, comfort with medical AI, and stigma history.
rundown: The study ran two scenario-based experiments: one with 300 participants comparing trust in a chatbot versus a human provider, and one with 450 participants choosing between them depending on wait time.

Measures covered trust in the provider, including perceived accuracy and bias, plus past experiences of medical stigma and comfort with medical AI. The authors conclude that preferences for AI-driven care are conditional.

For practice, the authors note that providers face high caseloads and appointment allocation challenges. They suggest that chatbots in care pathways could improve access and trust among marginalized populations for low-stakes concerns, as cost-effective, patient-centered options in selected contexts.
sources:
- peer_reviewed | Health Care Management Review | https://doi.org/10.1097/hmr.0000000000000513 | 2026-11-03
prev: b10cfddd8659e004639d6d43244fa60308535d4d17d5e0e0af5a289afbc00563

sha256

c651d5036709b5587e89aa30760bd0601f3c69acb7f1f42cdf11af33347c5eaf

previous

b10cfddd8659e004639d6d43244fa60308535d4d17d5e0e0af5a289afbc00563