Consent-GPT: is it ethical to delegate procedural consent to conversational AI? In their article ‘Consent-GPT: is it ethical to delegate procedural consent to conversational AI?’, Allen _et al_ 1 explore the ethical complexities involved in handing over parts of the process of obtaining medical consent to conversational Artificial Intelligence (AI) systems, that is, AI-driven large language models (LLMs) trained to interact with patients to inform them about upcoming medical procedures and assist in the process of obtaining informed consent. 1 They (...) focus specifically on challenges related to accuracy (4–5), trust (5), privacy (5), click-through consent (5) and responsibility (5–6), alongside some pragmatic considerations (6). While the authors competently navigate these critical issues and present several key perspectives, we posit that their discussion on trust in what they refer to as ‘Consent-GPT’ significantly underestimates one vital factor: the interpersonal aspect of trust. Admittedly, this interpersonal aspect is... (shrink)

Parity of esteemdescribes an aspiration to see mental health valued as much as physical. Proponents point to poorer funding of mental health services, greater stigma and poorer physical health for those with mental illness. Stubborn persistence of such disparities suggests a need to do more than stipulate ethical and legal obligations toward justice or fairness. Here, I propose that we should rely more on our legal obligations toward informed consent. The latter requires clinicians to disclose information about risks in a (...) way that is sufficient to satisfy what a prudent patient would reasonably want to understand in their circumstances. I argue that inadequate disclosure of the mental health complications of common surgeries risks exposing the craft specialists performing them to clinical negligence claims. Patients could argue they were counselled about said risks, improperly or not at all: improperly, if advised by a craft specialist lacking sufficient expertise in mental health; not at all, if mental health complications were simply forgotten. From this, I argue that a prudent approach for craft specialists would be to support and fund ‘integrative’ specialists (from rehabilitation medicine, liaison psychiatry and health psychology), more often to work alongside them within a multidisciplinary team that is better placed to navigate consent (via a prehabilitation process, for example). Based on duties toward consent, the extension of this type of coworking is another way to improve the resource and understanding accorded to mental health—but by starting within the citadels of physical health. (shrink)

When making substituted judgments for incapacitated patients, surrogates often struggle to guess what the patient would want if they had capacity. Surrogates may also agonize over having the (sole) responsibility of making such a determination. To address such concerns, a Patient Preference Predictor (PPP) has been proposed that would use an algorithm to infer the treatment preferences of individual patients from population-level data about the known preferences of people with similar demographic characteristics. However, critics have suggested that even if such (...) a PPP were more accurate, on average, than human surrogates in identifying patient preferences, the proposed algorithm would nevertheless fail to respect the patient’s (former) autonomy since it draws on the ‘wrong’ kind of data: namely, data that are not specific to the individual patient and which therefore may not reflect their actual values, or their reasons for having the preferences they do. Taking such criticisms on board, we here propose a new approach: the Personalized Patient Preference Predictor (P4). The P4 is based on recent advances in machine learning, which allow technologies including large language models to be more cheaply and efficiently ‘fine-tuned’ on person-specific data. The P4, unlike the PPP, would be able to infer an individual patient’s preferences from material (e.g., prior treatment decisions) that is in fact specific to them. Thus, we argue, in addition to being potentially more accurate at the individual level than the previously proposed PPP, the predictions of a P4 would also more directly reflect each patient’s own reasons and values. In this article, we review recent discoveries in artificial intelligence research that suggest a P4 is technically feasible, and argue that, if it is developed and appropriately deployed, it should assuage some of the main autonomy-based concerns of critics of the original PPP. We then consider various objections to our proposal and offer some tentative replies. (shrink)