invited · 45 min

Making sense of AI in clinical practice

After this session you will be able to say what your patients' use of AI changes in the consultation — and what to do about it on Monday.

atocp · boa annual congress · london   2026-09-24

In head-to-head studies, AI diagnosed more accurately than doctors, gave better-rated moral advice than an expert ethicist, and answered patients' questions in replies clinicians rated more empathetic than doctors'.

That was 2023 and 2024.

Tu et al., 2024 · Ayers et al., 2023 · Wilbanks et al., 2024

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

When technology gets in the way

There are many reasons for slow adoption of the stethoscope — among which are lack of formal education, hesitancy of the patient and physician to have an instrument placed between them, and lack of opportunities for continuing education after leaving medical school.

The stethoscope was resisted not because clinicians were ignorant but because it placed something between the clinician and the patient.

Reinhart, 2020

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Frontier AI capabilities

  • Long context: reason across very large bodies of text
  • Multimodal: read scans, figures, audio and video
  • Deep research: synthesise evidence from many sources
  • Tool use: search, run code, read PDFs, build prototypes
  • Domain reasoning: postgraduate level in most disciplines
  • Agents: work autonomously for hours; task length doubling every ~7 months

Kwa et al., 2025

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

AI in clinical practice

  • Diagnosis: matching or exceeding specialists across clinical domains
  • Expert–novice gap: narrows it, whatever the user's experience
  • Errors: AI errors and human errors are different errors
  • Communication: replies rated more helpful and more empathetic than physicians'

Tu et al., 2024 · Dell'Acqua et al., 2023 · Lenskjold et al., 2023 · Ayers et al., 2023

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

AI in physiotherapy

  • MSK reasoning: comparable to specialists on clinical reasoning tasks
  • Movement analysis: gait classified from mobile phone video
  • Documentation: AI scribes add 24% more time with the patient
  • Rehabilitation: programmes personalised from patient history
  • Remote care: AI-guided physio with ~85% patient satisfaction

Hao et al., 2025 · Reddy, 2025 · GOSH, 2025 · Michou, 2026 · Flok Health

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Sanctuary strategies

  • Denial: "It can't replicate the therapeutic relationship." Ignores the evidence
  • Retreat: "Focus on what only humans can do." A shrinking island
  • Restriction: "Don't let clinicians use it." They will, because it helps
  • Resignation: "Our roles will diminish." Helplessness dressed as realism

All four position you as a spectator of something happening to you.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

The question isn't whether AI changes practice. It already has. It's whether you shape that change.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

What AI systems lack

  • Stateless: no memory between sessions; you are a stranger every time
  • Static: knowledge frozen at training; no awareness of your field since
  • Contextless: no understanding of your expertise, patient population or reasoning patterns
  • Context is the bottleneck: model intelligence may matter less than context
§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Context sovereignty

You control your professional context, and use it to shape what AI does for you.

  • Your CPD, specialisms and interests, structured for AI
  • Context that accumulates as you work
  • Portable across any AI provider

What value do I bring? What value will I always bring?

Rowe & Lynch, 2025

§03 · what you bring

You bring context that AI doesn't have

your specialism your patient population local services and pathways your CPD priorities how you reason what you've seen work

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

What are agents?

  • Memory: retains context across time — your history, preferences, clinical patterns, previous decisions
  • Planning: sets goals and sequences actions, not just responding to prompts
  • Action: retrieves records, drafts notes, books follow-ups
  • Reflection: evaluates its own outputs and adjusts

From a tool you use to a system acting on your behalf.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Distributed cognition

  • Clinical reasoning has never been a solo act
  • It is spread across people, artefacts and environments
  • Your skill is managing that system: what to trust, how to reconcile conflicts, owning the synthesis
  • AI changes the system's composition, not its nature
§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

"Knowing things about conditions will become less important than knowing when to trust the outputs of AI systems."

Rowe, Nicholls & Shaw, 2022

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Professional context today

  • What does an AI know about your specialist area and clinical interests?
  • What does it know about the patient populations you work with?
  • What does it know about your CPD trajectory and the gaps you're working on?
  • What would it need to know to give you genuinely useful clinical support rather than generic output?
§04 · agency on both sides

Context sovereignty applies to your patients too

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Patients are already using AI

  • Patients use AI to research diagnoses, interpret symptoms, prepare questions and seek second opinions — before the appointment
  • Some arrive having already formed a view of their situation
  • The instinctive clinical response treats this as a problem to manage rather than an opportunity

These are patients exercising agency over their own health.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

From passive observation to active support

Conventional: react when patients arrive with AI-generated information.

Emerging: support better AI use before the encounter.

Before your next appointment, ask your AI: what questions should I be asking about my progress with [condition]? What should I understand about the next phase of treatment?

This is health literacy for the AI age.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Patient context sovereignty

Patients are building relationships with personal health AI that you may never see.

  • Their agents track history, symptoms, concerns and goals
  • That context accumulates over time, just as yours does
  • They may arrive knowing more about themselves than your clinical record does
§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

Multi-stakeholder relationships

The MDT is already a negotiation between professional contexts, expertise sets and institutional positions.

  • All stakeholders bring their own agents into the MDT
  • Each agent carries its principal's values and priorities into the encounter
  • The encounter becomes a negotiation between context-rich positions

This doesn't replace the therapeutic relationship. It changes what the relationship is.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

The professionals who will shape AI in clinical practice are those who bring deliberate context to it.

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

§05 · tomorrow, do this

Write down your professional context

Your specialisation, patient population, CPD priorities and how you reason — one page, explicit enough to give to an AI.

Give one patient a structured prompt

"Before your next appointment, ask your AI what questions you should be asking about your progress."

Bring judgement, not deference

Read the next AI output you meet as you would a colleague's opinion: engaged, contextual, accountable.

atocp · boa annual congress · 2026-09-24
questions

Thank you

mrowe@lincoln.ac.uk

QR code linking to the page for this talk, with the slides and handout
slides and further reading

atocp · boa annual congress · london   2026-09-24
§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

References

Ayers, J. W., Poliak, A., Dredze, M., et al. (2023). Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Internal Medicine. https://doi.org/10.1001/jamainternmed.2023.1838

Dell'Acqua, F., McFowland III, E., Mollick, E. R., et al. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality. SSRN Scholarly Paper No. 4573321. https://doi.org/10.2139/ssrn.4573321

Hao, J., et al. (2025). Large language models in musculoskeletal care: clinical reasoning and triage performance. Musculoskeletal Care. https://doi.org/10.1002/msc.70177

Kwa, T., et al. (2025). Measuring AI ability to complete long tasks. METR. https://arxiv.org/abs/2503.14499

Lenskjold, A., Nybing, J. U., Trampedach, C., et al. (2023). Should artificial intelligence have lower acceptable error rates than humans? BJR|Open, 5(1), 20220053. https://doi.org/10.1259/bjro.20220053

Great Ormond Street Hospital (2025). Ambient voice technology with generative AI (TORTUS): Phase 4 final report to NHS England. https://media.gosh.nhs.uk/documents/AAI_Phase_4_NHSE_Final_Report_1.3.pdf

Michou, E. (2026). Artificial intelligence in physiotherapy rehabilitation. Applied Sciences, 16(3), 1165. https://doi.org/10.3390/app16031165

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

References

Reddy, S. (2025). Classifying simulated gait impairments using privacy-preserving explainable AI and mobile phone videos. PLOS Digital Health. https://doi.org/10.1371/journal.pdig.0001004

Reinhart, R. A. (2020). The stethoscope in 19th-century American practice: Ideas, rhetoric, and eventual adoption. Canadian Bulletin of Medical History, 37(1), 50–87. https://doi.org/10.3138/cbmh.317-022019

Rowe, M., & Lynch, W. (2025). Context sovereignty for AI-supported learning: A human-centred approach. OSF Preprints. https://doi.org/10.31219/osf.io/8czva_v2

Rowe, M., Nicholls, D., & Shaw, J. (2022). How to replace a physiotherapist: Artificial intelligence and the redistribution of expertise. Physiotherapy Theory and Practice. https://doi.org/10.1080/09593985.2021.1934924

Tu, T., Palepu, A., Schaekermann, M., et al. (2024). Towards conversational diagnostic AI. arXiv. http://arxiv.org/abs/2401.05654

Wilbanks, D., Mondal, D., Tandon, N., & Gray, K. (2024). Large language models as moral experts? GPT-4o outperforms expert ethicist in providing moral guidance. PsyArXiv Preprints. https://doi.org/10.31234/osf.io/w7236

§01 what the tools do §02 the fears §03 what you bring §04 agency on both sides §05 tomorrow, do this

talk v1.0 final 2026-09-24
site michael-rowe.github.io/home-michael
orcid 0000-0002-1538-6052
licence CC BY 4.0 — reuse with attribution
cite as Rowe, M. (2026). Making sense of AI in clinical practice. ATOCP session, BOA Annual Congress, London.
/home/michael · build 1.0.0