Regulation and AI Scribes: How do we get the balance right?
This week, I co-authored an editorial in The BMJ with senior colleagues at Imperial College London, exploring the regulation of AI scribes in clinical practice. Our focus was on what regulation in this rapidly evolving area means for frontline clinicians and the delivery of patient care.
Regulatory Milestones
In recent months, there have been several key regulatory changes around the use of AI scribes in the UK:
MHRA guidance on AI as a medical device
NHS England guidance on deployment and governance
BMA position on adoption and meeting standards in general practice
Local Medical Committee guidance for general practices
Input from the NHS England CCIO on implementation considerations
Together, these documents reflect a growing consensus on the need for cautious but proactive regulation. But they also raise important questions for the clinical workforce.
The Work and Health Dimension
The conversation around AI scribes is about humans just as much as it is about technology.
AI scribes can improve clinical workflow, reducing documentation burden and freeing up time for quality patient interaction.
Working conditions in health and social care are at breaking point.
Better clinician working conditions often correlate with better patient care.
Clinicians are still required to “stay in the loop”, taking final responsibility for outputs generated by AI tools.
To make the benefits of AI scribes sustainable, healthcare systems must adapt how work is structured:
Clear role definitions when AI is used in clinical workflows
Defined AI supervision practices and support resources
Ongoing research to evaluate how AI affects service user outcomes, as well as how we think, work and practice
Regulation: Risk or Enabler?
Regulation is essential for safe implementation of new technology but it can also create barriers. In the UK, smaller healthcare organisations like GP practices may lack the capacity to manage complex governance or compliance requirements.
Good care looks different according to service user needs in different settings. There may be an advantage to a flexible approach to governance, where providers can tailor their implementation methods to best match their context while still meeting safe standards.
Clinician Perspectives on AI Scribes
I’ve been reflecting on how these developments apply to my own work and I’m not alone. Last week at a national conference, I asked a room of occupational health professionals where they stood on AI scribes. Out of nearly 60 respondents:
20% were already using an AI scribe
Over 10% planned to start soon
About 10% felt they had no need for one
Nearly 60% had not adopted AI scribes, citing various barriers
Of these, 25% pointed to regulatory uncertainty as the key obstacle
Others cited different barriers including cost and knowledge
Lara’s take
AI scribes are a case in point for how technology can transform how we work in healthcare. As with any tool, their value depends on how consciously we adopt them.
We should ask:
How can this technology improve working conditions and how do we carry this through to service delivery and service user outcomes?
How can we adopt this technology safely and sustainably, without creating inequality in implementation capacity?
Because of the pace of change, the choices we make now are forks in the road that may be difficult to revisit.
