Home » Insights » AI in Clinical Practice: Three Principles for Ethical Use

AI in Clinical Practice: Three Principles for Ethical Use

The rapid arrival of AI in clinical practice has raised urgent ethical questions, and a major professional body has now offered guidance. In a new position paper, the American College of Physicians (ACP) argues that AI use in medicine must be grounded in clinical integrity and professionalism (DeCamp et al., 2026). Although the paper is written for physicians, its principles translate directly to psychology.

Furthermore, the paper reframes AI as “augmented intelligence”, a tool that assists clinicians rather than replaces them. Below, we unpack the three guideposts it sets out and what they mean for South African practitioners.

Three guideposts for ethical AI in clinical practice

The ACP grounds its guidance in the patient-clinician relationship and sets out three guideposts: relationality, self-governance, and competence (DeCamp et al., 2026). Importantly, each is rooted in long-standing principles of medical ethics rather than in the technology itself.

Consequently, the framework is designed to outlast any particular tool. As AI systems change, the underlying commitments to relationship, judgement, and skill remain constant. Additionally, the paper stresses that clinicians and patients need practical guidance at the point of care, where privacy, disclosure, and fairness questions arise.

Relationality: protecting the therapeutic relationship

The first guidepost holds that AI should strengthen, not weaken, the relationship between clinician and patient (DeCamp et al., 2026). For psychology, where the therapeutic alliance is central to outcomes, this principle is especially important.

Specifically, greater automation should not reduce attention to communication, empathy, and shared decision-making. Ambient documentation tools, for example, may free clinicians to focus more fully on the person in front of them. However, technology should never become a barrier between practitioner and client.

Therefore, any AI tool you adopt should be judged partly on whether it supports genuine human connection. If it pulls your attention away from the client, its value is questionable.

Self-governance: keeping clinical judgement human

The second guidepost concerns independent judgement. As AI systems generate increasingly sophisticated recommendations, clinicians must retain the ability to assess whether an output is appropriate for an individual (DeCamp et al., 2026).

In other words, the ACP positions AI as an additional source of support, not an autonomous decision-maker. Moreover, the clinician remains accountable for the decision, regardless of what a tool suggests.

Ethics and professionalism sit at the heart of responsible practice.

Competence: guarding against deskilling

The third guidepost addresses a subtle risk. Maintaining clinical competence requires preventing AI-driven deskilling, so clinicians and trainees must retain enough expertise to evaluate AI outputs independently (DeCamp et al., 2026).

This matters because over-reliance can erode the very skills that allow you to catch a tool’s mistakes. Consequently, competence is not a one-time achievement but an ongoing responsibility. Trainees in particular must still develop core diagnostic and clinical reasoning skills.

The wider picture

The paper also raises equity and environmental concerns. Notably, it flags the energy demands of AI and the placement of data centres in vulnerable communities as issues of fairness. In addition, it calls for local governance that scales oversight to clinical risk and involves diverse stakeholders, including patients.

What AI in clinical practice means in South Africa

For South African psychologists, these principles map neatly onto HPCSA ethical guidelines around confidentiality, informed consent, and professional accountability. Therefore, adopting AI responsibly means applying familiar ethical reasoning to a new set of tools.

Practically, this means being transparent with clients about AI use, protecting their data, and never outsourcing your clinical judgement. Furthermore, it means choosing tools that support the relationship at the core of your work.

These themes connect closely to emerging thinking on responsible AI in mental health care. To keep your ethics knowledge sharp while earning CEUs, browse our 2026 CPD packages for South African psychology professionals.

References

Shopping Basket
Scroll to Top