A recent opinion cautions that the increasing use of artificial intelligence (AI) in healthcare — often labelled as "assistive" — could quietly transfer control over clinical information and decisions away from clinicians while leaving them legally and ethically responsible for patient care.
Different AI tools shift clinical work in different ways
The piece outlines how a variety of AI applications already in clinical environments affect the patient encounter. Examples include ambient documentation systems that convert recorded encounters into notes, predictive algorithms that generate risk scores within electronic health records (EHRs), and generative tools that draft correspondence such as insurance appeals, discharge summaries or patient instructions.
"The quieter and more likely one is that clinicians will continue to carry legal, ethical, and professional responsibility for decisions while gradually losing the time, information, authority, and discretion required to make them."
It argues that these technologies are not a single, uniform category: each tool reshapes workflows and raises distinct questions about consent, review and accountability. Of particular concern is a growing gap between where responsibility formally rests and where control and information actually reside.
Accountability, documentation and cognitive offloading
The opinion highlights several practical risks clinicians and health systems should consider:
- Ambient documentation can produce notes that omit the uncertainty or context a clinician would otherwise record, yet the clinician signs the note and is expected to assume its consequences.
- Predictive scores can channel attention to flagged risks while diverting scrutiny from other patient needs, especially when time pressure encourages clinicians to accept recommendations with a single click.
- Generative tools drafting appeals or instructions may be sent under a clinician’s name, raising questions about review processes and authorship.
These dynamics can lead to what the opinion describes as "cognitive offloading": clinicians delegating parts of reasoning, documentation and triage to algorithms because it saves time or reduces friction, even when oversight is desirable.
Implications for South African practice and regulation
While the commentary is framed in international terms, its concerns have direct resonance in South Africa. Local hospitals and clinics are increasingly adopting digital systems and telehealth, and international AI vendors often supply tools integrated into EHRs or deployed as add-on modules. If clinicians remain formally accountable for records and decisions while losing practical control, this could complicate medico-legal cases, professional disciplinary matters and patient safety investigations.
Regulators, professional bodies and health managers will need to clarify expectations for:
- who must review and sign AI-generated content and what minimum verification is required;
- how informed consent covers the use of ambient recording or AI-assisted documentation;
- audit trails that identify human versus machine contributions in clinical records.
The opinion calls for more granular thinking: not treating "AI use" as a single policy problem but distinguishing among documentation, prediction, generation and triage systems and tailoring governance accordingly.
Practical steps for clinicians and managers
For individual clinicians, the commentary suggests maintaining active review habits and resisting the temptation to accept AI outputs uncritically, especially under time pressure. For health leaders, the piece recommends clear institutional policies that define review responsibilities, documentation standards and escalation procedures when AI outputs are uncertain or discordant with clinical judgement.
As AI becomes more pervasive in hospitals and clinics, health professionals should expect regulators and professional councils to update guidance on documentation, informed consent and supervision. Patients and families also deserve transparent information about how their records are created and how AI may have contributed to care decisions.
Technology can augment clinical care when deployed thoughtfully; the central challenge is ensuring that it remains an aid to human judgement rather than a substitute for it. Clinicians concerned about how AI is used in their workplace should raise the issue with their hospital manager, professional body or regulatory authority. As always, individuals should consult their doctor or clinic about personal health concerns rather than relying on algorithmic outputs alone.
