The AI Scribe Study

We will investigate current, and evolving use, of AI Scribes in UK General Practice to (1) identify and understand the influence on quality of care for patients, and experiences of primary care staff, as well as ways in which it may cause or exacerbate inequalities, and explore other unintended consequences, and (2) develop recommendations and inform guidelines for the regulation, adoption and use of AI Scribes.

In a context where GPs are overwhelmed, the appeal of AI Scribes is clear. However, we know digital innovation can often have unintended consequences and there is an urgent need to look further than initial time-saving and satisfaction. Current evidence into the longer-term indirect influences, as well as patients’ perspectives, is scarce, and understandings in relation to how AI Scribes may affect all dimensions of care quality (safety, effectiveness, patient-centredness, timeliness, efficiency, equity) is limited. Once we understand these areas better, we can ensure optimal use of such tools for both patients and clinicians.  

The study starts with a literature review and stakeholder consultation, followed by two main data collection stages. Firstly, we will carry out observations of AI Scribe use in general practices, alongside interviews with patients and clinicians. Secondly, we will compare notes written by GPs with those generated by AI Scribe systems. GP and patient working groups will review the recordings, compare the AI Scribe summaries and GP notes, and discuss the differences. Finally, we will bring together findings in workshops and co-produce guidance on how to use AI Scribes safely, fairly, and in ways that do not compromise quality of care.

This study will provide novel insights into how AI Scribe use can influence the consultation, patient notes, and quality of care. We will explore how certain groups may benefit from, or be disadvantaged by, AI Scribe use. Guidance will be co-produced on how to use AI Scribes safely, fairly, and in ways that build trust. The findings will also inform how consent processes can be designed to be accessible and inclusive, so patients feel informed, respected, and confident about their care.

Putting findings into practice:

Every 6 months, knowledge mobilisation workshops will allow discussion of cumulative findings and integration of stakeholders’ perspectives. A policy maker round table will be held at the end of the study, to inform AI guidance and policy making decisions. Findings will be submitted to peer reviewed academic journals and relevant conferences, presented to key organisations and contribute to the ongoing development of AI Scribes guidelines produced by the NHS and other organisations such as the RCGP. In collaboration with our ‘people and communities’ panel we will develop accessible public facing outputs e.g. infographics, sharable videos and social media.

Patient and public involvement (PPI):

We ran two workshops with 8 public contributors from varied backgrounds to inform our study design, see summary here. There are two lay co-investigators on the study team who will co-lead our diverse ‘people and communities’ panel. Members of the panel will provide input throughout the study during regular panel meetings. Involvement includes (but is not limited to) 1) developing recruitment materials; 2) attending stakeholder workshops; 3) developing and piloting the interview schedule; 4) interpreting findings; 5) attending the working groups; 6) supporting inclusive, accessible dissemination.

Project timeline

The study commences in May 2026 and will be completed in June 2028