Artificial intelligence (AI) is already entering physiotherapy practice but efficiency alone is not a good enough reason to adopt it. 

An ethics lens can help physiotherapists judge where AI adds value, where it creates risk and what should remain firmly human-led. 

Why efficiency is not enough 

AI has moved from media hype into mainstream use across many domains. 

It can summarise, draft, transcribe and reorganise information and is increasingly marketed as a transformative productivity tool for roles that involve information work. 

That appeal is easy to understand in physiotherapy, where the burden of documentation, correspondence and administrative tasks is often discussed. 

However, physiotherapy should not rush to adopt AI simply because it is presented as an efficiency solution. 

AI may assist with some parts of practice but it can also be inaccurate, incomplete, overconfident or poorly matched to the clinical situation. 

More importantly, efficiency should not be prioritised above what physiotherapy most depends on: judgement, communication, patient relationships, context-sensitive decision-making and professional accountability. 

The key consideration is where AI genuinely adds value to physiotherapy practice and where that value may come with important trade-offs. 

Rather than starting with the tool and asking where it can be used, we should begin by asking what good physiotherapy looks like and then consider whether AI supports that vision. 

Healthcare has long used ethics to guide its understanding of best practice. 

In physiotherapy, principles such as beneficence, non-maleficence, autonomy and justice provide a practical framework for judging where AI may support care, where it may introduce risk and how trust should be protected. 

Doing good, not just saving time 

Beneficence is the ethical principle of doing good. In this context, that means asking how AI in physiotherapy might genuinely improve care, clarity or sustainability rather than simply making tasks faster. 

There may be value in using AI for some forms of information work. 

That could include assisting with administrative tasks, improving the structure of service documents, formatting standard patient education materials on topics such as falls prevention or self-management, or summarising evidence relevant to a focused clinical question. 

Small gains in these lowerrisk tasks may be worthwhile if they free up more time for clinical work. Polished output is not the same as reliable output, however. 

Any benefit is only real if the result is accurate, useful and worth the time needed to review it. Fake efficiency remains a risk. Beneficence therefore requires selective use of AI in situations where it adds real value rather than simply appearing efficient.

Avoiding harm in physiotherapy 

Non-maleficence is the ethical principle of avoiding harm. In physiotherapy, harm may arise not only through physical risk but through inaccurate advice, misleading summaries, missed contraindications, oversimplified explanations or confident wording that hides uncertainty. AI may help reduce some forms of harm. 

For example, it may assist with reviewing groups of related documents – such as patient handouts, service information or discharge resources – to identify inconsistencies, omissions or wording that could create confusion. 

But AI use requires caution. AI can sound authoritative while being wrong and it may miss context that an experienced physiotherapist would recognise immediately. 

AI may assist with drafting or review but clinical judgement and final responsibility must remain with the physiotherapist.

Supporting the patient’s voice 

Autonomy is about more than a person’s right to choose. 

It is about supporting informed and meaningful participation in decisions about care. 

In physiotherapy, that means not only explaining options but helping patients understand the rationale for them, apply them to their own life and values, weigh trade-offs and express what matters most in their particular situation. 

Physiotherapists could use AI to support autonomy by helping patients organise concerns, prepare questions before an appointment or better understand the likely purpose, burden or demands of a treatment plan. 

But AI can also undermine autonomy. It can create false certainty, substitute for professional explanation and shared decision-making or subtly steer decisions through framing that reflects bias, norms or what is common rather than what is right for the individual. 

AI supports autonomy only when professional judgement ensures that it is helping patients participate more fully in decisions rather than narrowing or directing choice. 

Fairness, access and care

Justice is the ethical principle of fairness. In physiotherapy, this includes whether patients can access physiotherapy expertise, understand the information provided and make use of it. 

AI may support justice in practical ways. It may help services create easier-to-use versions of standard patient resources such as easy-read formats, translated handouts or the same material presented as a checklist or short summary. 

This matters because patients have different languages, literacy skills and levels of confidence or energy to work through standard health information. AI may also reduce time spent on routine drafting or administrative work. 

That time saved could be used to see patients who might otherwise miss out because of waitlists, workforce pressure, geography or service model constraints. 

However, time saved does not automatically lead to fairer access. 

It matters how those gains are used. If they mainly make care more convenient for people who already have access to physiotherapy or allow those same patients to use more of the service, AI may entrench existing inequities rather than reduce them. 

Justice requires deliberate decisions about how time savings are spent.

They should help reduce barriers and extend fairer access rather than simply improving convenience for those already well served. 

Trust, safeguards and professional judgement 

All of these ethical principles depend on trust: trust that physiotherapists will use information wisely, communicate carefully and apply professional judgement in ways that serve patient care. 

Protecting privacy through secure systems is part of that ethical responsibility and should be treated as a minimum standard wherever AI is used. 

But security is not the whole issue. Some purpose-built note tools now do more than turn brief notes or recordings into more complete records. They also suggest what has been omitted, potential interventions or patient education resources. 

The issue is not whether these tools are helpful. 

It is whether, over time, they begin to shape what clinicians notice, record or prioritise in ways that seem reasonable on the surface but may subtly distort the clinical picture by overemphasising some issues while underrecognising others. 

Patients may lose confidence if AI appears to be replacing attention, skill or judgement rather than supporting them. 

Safe use therefore depends on clear boundaries, careful oversight, transparency and retained human accountability. 

What this means for practice 

AI is likely to have an increasing presence in physiotherapy but that presence should enhance our ability to care for patients and maintain the trust of our communities. 

The same ethical standard should apply across every use, with safeguards that match the level of risk. Privacy, security, clinical review and professional accountability must be assured for any AI use. 

Physiotherapists should not adopt a mindset of using AI wherever possible simply because it appears to save time or improve efficiency. Some efficiencies will be false and others may come with significant trade-offs. 

We should start with a vision of good physiotherapy, guided by ethics, and then ask where AI might genuinely support that work. AI holds real promise for physiotherapy when it strengthens care without weakening judgement, fairness or trust. 

>>Pete Haynes is a physiotherapist at Mercy Health in Albury, New South Wales, who writes about technology and healthcare. He also builds digital tools for physiotherapists, including ExerciseScript and Physiflow, with a focus on practical innovation in clinical practice.