Speaker 1:
Dr. Jared Powell:
Hello, it's Jared. There's a conversation happening right now across physiotherapy and really across all of healthcare that I think deserves more scrutiny than it might be publicly getting. It goes something like this. AI is coming. AI can explain a diagnosis to a patient. AI can write an exercise program for a patient. AI can summarize the latest systematic review and meta-analysis faster than any of us ever could. It's superhuman, and therefore also the argument goes, we need to lean into the art of what we do, the human side, the connection, the skill, the things that a machine can't replicate.
Dr. Jared Powell:
I understand the appeal of this argument. It feels right. It feels protective. It gives us something to hold onto in a moment where the ground feels like it's shifting beneath us, but I think it's the wrong lesson. I don't think it's the wrong lesson because the human dimension doesn't matter quite the opposite. The human dimension is absolutely pivotal. I think it's wrong because this line of reasoning smuggles in a set of assumptions that I think we need to challenge before we build a professional identity around them. And the biggest assumption is this, that science and art are separable components of clinical practice and that AI neatly replaces one while leaving the other untouched. I don't think that's true, and in this episode I want to explain why now this debate isn't new. The art versus science question has been rumbling through physiotherapy for as long as I've been a physiotherapist for 15 years or more well before AI entered the conversation.
Dr. Jared Powell:
In fact, I wrote a blog post about it some years ago, sparked by a social media post defending manual therapy from the relentless criticism that seems to receive the comment section were pretty predictable. Emotional justifications for interventions with minimal evidence, passive aggressive swipes at younger clinicians who dared to ask for I rationale appeals to decades of clinical experience as if time served is the same thing as actual expertise, which as I argued last episode, it isn't On one side you have the old guard. The physiotherapy is an art form. You can't measure everything. You can't reduce what I do to a randomized control trial, and to be honest, there is something in that Being an expert, having decades of clinical experience is valuable. There is something in that indeed, not everything that matters in a clinical encounter is quantifiable, but this position has also been used to historically to deflect legitimate scrutiny.
Dr. Jared Powell:
The proclamation, it's an art can very quickly become, don't question me, don't question my clinical practice. Your randomized control trial tells me nothing about how I practice, and I think we should be wary of that line of inquiry. On the other side, you have the so-called new guard, the evidence-based crowd. Show me the data, show me the systematic review. Show me the RCT. If you can't cite a meta-analysis, you are practicing faith not physiotherapy. And again, there's something in that too. We should absolutely be anchored in evidence, but this position has its own shortcomings. It can become quite rigid, formulaic, and detached from the person sitting in front of you. Indeed, no randomized control trial tells you how this individual facing you is going to respond to a particular treatment quoting a number needed to treat to someone who is terrified of their MRI result is technically evidence-based and practically useless.
Dr. Jared Powell:
The bio-psychosocial model, which this camp loves to champion as do I, has been bastardized and misinterpreted more times than I can count. It's often reduced to a slogan, a heuristic rather than being genuinely applied. So we've had two camps for decades. Art versus science, intuition versus evidence, hands on versus data, and neither camp has been entirely wrong, but both have been incomplete. AI has just given this old argument, a new accelerant. Let's start with the claim that AI replaces the science. What do most clinicians actually mean when they say the science in everyday practice? What I've come to notice is they usually mean things like looking up a clinical guideline, recalling prevalence data for a certain musculoskeletal condition following a diagnostic algorithm or protocol, prescribing a textbook, ex exercise progression, pulling the conclusion from a systematic review and delivering it to a patient. And yes, AI is very good at this. In many cases, it's faster, it's more comprehensive and more up to date than any individual clinician could possibly be. It doesn't get tired at four o'clock on a Friday afternoon. It doesn't need a coffee. It doesn't have a bad sleep. It doesn't forget the inclusion criteria of that randomized control trial that you read six months ago, but none of that is science, not in any meaningful sense as far as I'm concerned.
Dr. Jared Powell:
That's information retrieval, that's protocol following, that's pattern matching, and if that's what your science looks like in the clinic, then sure, AI is absolutely going to threaten it. However, I believe the threat isn't that AI is doing science. The threat is that we were calling something science that never really was. Now, I've talked about the philosopher Carl Popper before, and I won't rehash the whole argument from last episode, but the short version is this science, real proper science is conjecture and refutation forming a hypothesis, testing it against reality in the form of an experiment and revising when it fails. That process is creative, it's contextual, it's deeply human, and it is exactly what good clinical reasoning looks like when it's done properly. What AI replaces isn't the science, it's the impression of science. It's the facade of science. It replaces fake science, the performative version of science, the version where a clinician googles a systematic review in the treatment room and reads the conclusion out loud.
Dr. Jared Powell:
That has never been science and it will never be science. It's simply information delivery dressed up as a performative version of science. Okay, now let's look at the other side, the call to lean in to the art. I have concerns here too, and I've already flagged the historical baggage, but I think there is also a deeper issue when people say lean into the art in response to the rise of ai. I think what this often means is lean into the ineff stuff, the intangible, the human touch, the things you cannot put into words, and that sounds really nice, but again, it's imprecise and I think imprecision in this area has real consequences because if you can't articulate what the art actually is, you can't teach it, you can't assess it and you can't improve it. It just becomes a feeling. And art without rigor, I would argue these just vibes and vibes don't hold up.
Dr. Jared Powell:
When someone asks you why your patient isn't getting better after 12 weeks, vibes, most certainly do not hold up in a court of law, and they certainly don't hold up against a well-designed chatbot that can deliver information more clearly and more consistently than a burnt out clinician running 15 minutes late. So I don't think lean into the art is a sufficient response to the rising tide of ai, but I also don't think the art crowd is pointing at nothing. I see. I think there is something valid here. They're pointing at something real. I just don't think they have the right language for it, and I want to offer some better language today. The first thinker I want to bring in is Michael Palani, a renowned polymath who worked across physical chemistry, economics, and philosophy. Now, he wrote extensively about what he called tacit knowledge. His most famous line is we know more than we can tell, and I think that quote captures something essential about clinical expertise that the art versus science debate completely misses.
Dr. Jared Powell:
Think about what an experienced clinician picks up in an encounter that never makes it into the clinical notes, the way a patient hesitates before answering a question about their sleep, A subtle shift in their posture when you mention returning to work, the quality of their movement when they don't think you're watching versus the guarded effortful version they perform when they know you're assessing them. The tone in the voice that tells you someone is frightened, even though their words say they're fine. Now, none of this is mystical. None of it is art in the romantic or aesthetic sense. It is knowledge, but it's a kind of knowledge that resists being made fully explicit. Unfortunately, it can't be written into a guideline. It can't be captured in a large data set, and it certainly can't be accessed by an AI system that processes text and has no body no felt sense of what it's like to be in a room with another person who is suffering.
Dr. Jared Powell:
Now, Palani argued that this tacit dimension is present in all skilled performance from riding a bicycle to making a medical diagnosis. The expert doesn't seem to follow rules. Indeed, they've internalized something deeper than rules. They tend to perceive patterns and respond to them in ways they may not be able to fully articulate, but I don't think that makes those responses irrational or unscientific or simply artistic. It makes them a form of knowledge that lives in a different dimension or a different register. This I think is what the art crowd is actually pointing out. Art as tacit knowledge, embodied, situated experiential knowledge that accumulates through years of attentive and reflective practice cannot be fully formalized to explicit rules or algorithms. I think this is at the heart of what being an expert clinician really is. The second thinker I want to introduce is Donald Shern, who wrote a brilliant book called The Reflective Practitioner.
Dr. Jared Powell:
Shern made a distinction that I think every clinician should know about. He talked about the high ground of technical rationality and the swampy low lens of real practice. The high ground is where problems are clean and well-defined. There's a clear diagnosis, there's a clear guideline to follow. There's a clear inclusion criteria. The evidence is strong, the intervention is straightforward. This again, is the territory that AI occupies beautifully. Give it a clean problem with a clear answer and it will outperform most clinicians, myself included. However, as you well know, swampy lowlands is where clinical practice actually lives. The patient with three comorbidities in a complex social situation who doesn't neatly fit in any clinical category, the person who doesn't match the imaging, whose story keeps shifting the patient who has sadly been everywhere seen everyone tried everything valiantly remains sitting in front of you the last resort, you can see the desperation in their eyes that you are going to be the person that helps them.
Dr. Jared Powell:
The patient who technically has a straightforward clinical presentation whose fear beliefs, the real barrier, sadly no protocol can address. Sharon's argument was that professional expertise lives in the swamp, not on the technical high ground. I think he's absolutely right. The high ground is where protocols work. The swamp is where thinking is required. If AI handles the high ground brilliantly, that doesn't diminish the clinician. It clarifies where the clinician's real value lies. It's in the swamp, it's in the mess, it's in the chaos. It's in that everyday struggle, complexity of an individual patient. It's where the algorithm runs out and you have to reason your way forward with another human being. The third idea is from Aristotle. Yes, that Aristotle godfather of philosophy and it's one of the oldest ideas in philosophy, but I think it's never been more relevant. 2000 years later, Aristotle made a distinction between three types of knowledge quickly, EPIs, which is the theoretical or scientific knowledge, knowing that something is the case. TechNet, which is technical skill or craft, knowing how to produce something, and then resis, which is practical wisdom or practical knowledge, knowing what the right thing to do is in a particular situation. Now, this is where it gets important. The art versus science debate and physiotherapy maps almost perfectly onto EPIs versus TechNet. The science crowd champions EPIs, show me the evidence, show me the data, show me the theoretical framework.
Dr. Jared Powell:
Conversely, the art crowd champions TechNet. I have a skilled pair of hands, which what I can do with these skilled pair of hands, I can implement a precise manual therapy intervention. It's going to shift L four on L five that's going to eradicate an obstruction in this person's joint, which is going to fix their back pain and in some cases fix their cough. But interestingly, neither of these crowds talk about essis and essis, I argue, is the thing that actually makes a clinician good and useful esis is the capacity to judge knowing when to push and when to back off, when to explain, and when to just shut up and listen and the evidence applies. And when the person in front of you is the exception to that rule or the exception that proves the rule. When a patient needs information and when they simply just need permission to move, it is in a sense the integration of knowledge and context and relationship in that clinical coalface moment.
Dr. Jared Powell:
I don't think it can be reduced to simple third person science. I don't think it can simply be reduced to an art form. It's a third thing. It's another thing altogether, and it's I think the thing that AI fundamentally cannot do because essis requires being embedded in a situation with a person, with stakes, with uncertainty and exercising judgment that is irreducibly contextual. I think if our profession adopted the language of essis, and maybe we can come up with another term for it or of practical wisdom, we could finally put the art versus science debate to rest because it was never really about art or science. It's always been about our wisdom. The fourth and final thinker I want to bring in is Herbert Dreyfus, who was a philosopher at Berkeley and one of the earliest and most serious critics of artificial intelligence, and he makes his arguments In the 1970s, long before chat GPT, long before Claude, long before Gemini, long before anyone was worrying about AI replacing clinicians, and if we listen to what he said 50 years ago, his ideas were eerily prescient.
Dr. Jared Powell:
Dreyfus argued that AI would succeed at replacing tasks that can be formalized into explicit rules, but fail at tasks that require embodied situated expertise. His point was that human skill at the expert level is not simply rule following. It's intuitive in the philosophical sense. It's not a gut feeling, it's not an uneducated guess, but it's the product of a deep experience that has been internalized to the point where the expert perceives and responds to situations holistically rather than analytically. It's like the chess Grand Master doesn't calculate every possible move using probability theorem. They see the board, they can intuit the next move. Similarly, the experienced clinician doesn't run through a mental checklist of differential diagnoses. They seem to recognize a pattern and respond appropriately, and this recognition is embodied, it's situated, it lives in the encounter. It doesn't live in a textbook. It doesn't live in a large language model.
Dr. Jared Powell:
Dreyfus predicted decades ago that AI would replace this kind of expertise, and I think in many ways he was right. Large language models are spectacularly good at information retrieval, text generation and pattern matching across massive data sets. They're not good at being in a room with a frightened human being and knowing what to say next. They fundamentally lack what DRA has called the embodied coping that defines genuine expertise. So the long and short of it, this doesn't mean that AI is useless far from it. AI is useful in a specific domain. The explicit, the formalized, the codifiable and clinical expertise, the kind that actually changes outcomes operates in a different register or different dimension entirely. One that is tacit, contextual, embodied and relational. AI can inform a patient, it can retrieve information for a patient, but it most certainly cannot think with them, or at least I don't think so, and certainly not at the moment.
Dr. Jared Powell:
So where does this leave us? I think the art versus science debate was always a false dichotomy, and I think AI has made the falseness of that dichotomy impossible to ignore. What AI does is hold up a mirror that reveals our deepest darkest flaws. And in the case of AI and physiotherapy, what that mirror shows is how much of our practice was neither art nor science. It was information delivery, it was information retrieval, it was rule following, it was protocol following. I think it was probably routine dressed up as expertise, and I think if we we analyze that and we sit with that, that's confronting. But it can also be clarifying because once you strip away the parts of clinical practice that AI can do better, what remains is something specific and nameable, it's tacit knowledge. In pal sense, the things we know but cannot fully articulate.
Dr. Jared Powell:
It's reflective practice in the swampy low lands. In Shen's case, it's the ability to reason through messy ill-defined problems that no algorithm can solve. It's esis in Aristotle sense, practical wisdom, the judgment to know what the right thing to do is in this specific clinical scenario with this person in this context. And it's also embodied expertise in dreyfus sense, situated experiential and irreducible to a set of explicit rules. This is something more integrated than either an art label or a science label captures, and if I had to put it in a single sentence, I'd say it this way. The work is thinking with another person under conditions of uncertainty. That was true before ai. It will be true long after ai. AI hasn't changed the nature of our work. I think it's just made the nature of the work harder to ignore. So what do we do with all this?
Dr. Jared Powell:
Practically? What are the take homes? I think we can use AI for what it's good for, information retrieval, summarizing evidence, generating patient resources, reducing our administrative burden. Let it handle the high ground. Free yourself for the swamp because patients need you for the swamp, but I think we should invest relentlessly in the thing that AI cannot replicate. That's your clinical reasoning, that's your capacity, your humanity to sit with uncertainty rather than rushing to false certainty. Your ability to notice what a patient isn't saying, your ability to notice their movements, their posture, their uneasiness, when you bring up a certain question, this is your practical wisdom. These aren't innate gifts. They are cultivated and they also seem to compound over time. But only if you practice reflectively, not just repetitively. 10 years of thoughtful practice builds esis 10 years of repeating the same thinking does not, and for the love of the profession, I think we should stop framing this whole debate as art versus science.
Dr. Jared Powell:
That debate has run its course the game's over. The real question was never whether physiotherapy is an art or a science. I think the real question has always been whether you are doing the kind of work that requires genuine expertise or the kind that a chat bot could handle, and if the answer is uncomfortable, good, that discomfort is where the growth is going to be. We are knowledge workers in service of people. Again, I said this in the last episode, but this episode, I want to expand on that. I want to add something different. I want to bring another layer to it. We are specific kind of knowledge workers. We are practitioners of resis or wisdom of practical wisdom in the service of people who are suffering and uncertain. This is something I think worth building an identity around, building a profession around something worth defending and something that ultimately no machine, however sophisticated is going to take from us unless we give it away by refusing to do the hard work of earning it. I'm Dr. Jerry Powell, and this has been another episode of The Shoulder Physio Podcast. I'll see you next time. The Shoulder Physio podcast would like to acknowledge that this episode was recorded from the lands of the terrible Lang people. I also acknowledge the traditional custodians of the lands on which each of you are living, learning, and working from every day. I pay my respects to elders past, present, and emerging, and celebrate the diversity of Aboriginal and Torres Strait Islander peoples and their ongoing cultures and connections to the lands and waters of Australia.
