On Acupuncture, A.I. and Accompaniment

Should acupuncturists worry about being replaced by robots?

In response to the post about community acupuncturists as cyborgs, an alert subscriber pointed out that since we’re talking about machine intelligence, did I know that the NCCAOM has a new taskforce on A.I.? So I went down that rabbit hole.

The NCCAOM claims that “A.I. is set to revolutionize acupuncture”. That’s a sweeping pronouncement. But what does it mean, exactly?¹ Since nobody’s suggesting that robots should needle acupuncture points on humans (not yet anyway), the conversation about A.I. and acupuncture seems to be about its role in diagnosis and prescription.

According to an article in Acupuncture Today,

AI-driven algorithms can compile a patient’s medical history, lifestyle factors, and past treatment responses to recommend optimal acupuncture points and protocols. Some innovative startups are already developing AI-powered diagnostic tools that use machine learning to assist acupuncturists in clinical decision-making. This technology enhances accuracy, improves treatment consistency, and helps bridge gaps in diagnostic interpretation, which has long been a challenge in the field…studies highlight AI’s potential in traditional medicine, demonstrating how these tools can streamline diagnostics and treatment strategies. The ability to identify patterns in a patient’s condition – based on an analysis of thousands of similar cases – leads to more precise, evidence-based care.

Huh. I can think of a few problems with the idea of A.I. delivering “precise, evidence-based care” and “optimal acupuncture points and protocols”. In no particular order:

I know I keep bringing this up, but — to the best of my knowledge, somebody please correct me if I’m wrong — there’s no robust research on comparative efficacy of different kinds of acupuncture (which are extremely variable!) Broadly speaking, we’ve got no real evidence as to whether TCM acupuncture works better than Balance Method or Korean 4 Point or Jingei or Master Tung or Five Element — or literally any other clinical approach. Acupuncturists love to argue over this stuff but it all seems to come down to “my teacher said” or “I like doing it this way” rather than research. How can we have evidence-based care when we’ve got no evidence base to address comparative efficacy? How would you even get to “optimal”?

Maybe the author is referencing Traditional Chinese Medicine research and, as people in the U.S. often do, assuming that TCM = acupuncture? (It doesn’t.) Even so, I’d like to see the research on comparative efficacy of “optimal” acupuncture point protocols even within TCM. There would have to be a lot of comparative studies on a lot of different protocols to adequately address the diversity of issues we see in community acupuncture. I’d be surprised if that volume of comparative research on point protocols…exists? But if I’m wrong, I’d also love to see it because it’d be fascinating.

The usefulness of A.I. tools depends on the quality of information that’s fed into them. I have a lot of questions about what exactly we’d be feeding our new toolbox. So if any acupuncture A.I. enthusiasts are reading this, please point me in the direction of your evidence base.

Art by A. Cohen

Acupuncture points by their very nature aren’t “precise"; acupuncture’s a shotgun, not a laser. Unfortunately, acupuncture treatments don’t deliver clinical results in proportion to the “accuracy” of a diagnosis. (Lots more about that here.) Any acupuncturist who’s given a lot of treatments recognizes the subtle and nuanced impact of acupuncture points, especially the synergy of points acting together. That’s what we like about acupuncture, its subtlety and its nuance.

This seems like a good opportunity to revisit a conversation from a year ago with Dilip Babu, M.D., who’s a POCA Tech Board member and a WCA acupunk. We were talking about diagnosis in acupuncture vs. biomedicine and he said:

Allopathic medicine understands illness in terms of diagnostic categories or disease entities that sort of exist in the ether, separate from a specific human being. Whereas Ayurveda and acupuncture look at illness and health in a more embodied way. Like, the way you as an individual might embody an imbalance is different than one of these abstract biomedical disease entities.

And then the diagnostic categories get turned into billing codes. It’s a very short road from the philosophy of allopathic medicine to the economic reality. If you don’t have a billing code for it, it doesn’t exist. There’s this idea of disease as a thing separate from people, and then it’s imposed on the individual patient.

Sometimes the abstract category fits what the patient is experiencing pretty well, so you take it from there and proceed with the “standard of care” treatment pathway. The problem is that sometimes the abstract category doesn’t quite fit, in which case you might end up blaming the patient. And if things really aren’t working, that’s when patients get labeled “noncompliant” or some other stigmatizing language like that.

I was trained in a TCM acupuncture school, and my experience in clinic was absolutely about trying to fit real humans into the abstract categories of TCM syndromes. I understand this isn’t necessarily how TCM is supposed to be practiced, but often in the US, that’s what happens. (And yes, it makes more sense with Chinese herbal formulas, but I’m not an herbalist and herbs aren’t acupuncture.) Outside of TCM, diagnosis in acupuncture is often about identifying stagnation in a channel. How can that possibly be evidence-based, when the diagnostic process is so individual and subjective (for example, palpation) — not to mention that we can’t even prove the existence of the channels themselves?

For better or worse, a lot of acupuncture theory — including acupuncture diagnosis — is applied metaphor.² Many humans struggle with that, let alone A.I.s. Which gets us to the issue of clinical judgement. What does optimal acupuncture care actually look like in a clinical setting?

Something I’ve learned from working in community acupuncture for many years: the reason people are suffering usually isn’t because they lack the right answers. Chronic illness is, well, chronic and even the “correct” acupuncture diagnosis only gets you so far. Factor in the social determinants of health and you’ll see stubborn conditions get even more stubborn. With acupuncture, providing relief is usually less about coming up with revelatory answers and more about building relationships (for big clinics like WCA, that includes not only patients’ relationships with individual practitioners but with the clinic itself).

With acupuncture, optimal care only happens in the context of relationships.

I’ve seen a lot of smart people get very disappointed with the clinical practice of acupuncture because they graduated from school with the expectation that patients were puzzles to be solved as opposed to humans to care for. If you like puzzles, acupuncture might sound like a satisfying career (leaving aside, for the moment, the issue of pay). The acupuncture establishment contributes to this impression — especially the NCCAOM, whose exams define both entry-level competence and acupuncture safety as equivalent to solving the puzzles of TCM diagnosis.

In community acupuncture, though, optimal care is a function of accompaniment — a concept that Dr. Paul Farmer borrowed from Liberation Theology and applied to public health: “(E)specially in the most urgent circumstances and often clinical deserts, the moral responsibility of the healer is to step inside patients' experiences and accompany them through the worst moments with empathy and expertise, compassion and care, for as long as it takes.” Accompaniment recognizes that healing isn’t just a function of what you know, it’s what you do. It’s relational; it’s praxis.

Accompaniment, like other concepts drawn from liberation studies, is about humanization — often in the face of dehumanization³. How can we feel more human, more like subjects and less like objects? Or like abstract disease entities? What people appreciate about community acupuncture — both patients and practitioners — is that it helps us feel human, together.

In the context of accompaniment, clinical judgement means discerning how best to care for the actual people who show up in our actual clinics. To repeat, marginalized people’s needs aren’t concepts. Meeting them requires sensitivity, empathy, intuition and flexibility. Not to mention, reading body language and knowing your way around trauma-informed care — dozens of tiny moment-to-moment decisions that are part of treating someone in a community clinic. For example, do you ask someone to roll up their sleeves and take off their socks, or do you work with what they give you, which might be just hands, ears and scalp? It depends on someone’s expression, their posture in the chair, how you feel when you’re sitting next to them on your rolling stool: it’s all subtlety and nuance, all the time.

It’s hard to imagine how A.I. tools could assist with any of that, let alone “revolutionize” it.

Art by A. Cohen

Please forgive me for detouring into acupuncture politics — but what this conversation about A.I. reminds me of — a lot — is some unfortunate ideas about “tiered practice” that I’ve heard from certain acupuncturists. The acupuncture profession has a tortured relationship with its own degrees (particularly their titles, which is a whole other thing). Back when we started POCA Tech, the narrative in the acupuncture profession was that Master’s degrees would soon disappear altogether, to be replaced by entry-level doctorates. That didn’t happen, of course, and now the doctoral degrees are looking more and more like a millstone around the acupuncture profession’s neck.⁴

Nonetheless I still sometimes meet acupuncturists who think that community acupuncturists with Master’s level training should be supervised by Doctors of Acupuncture. Regardless of the fact that we all hold the same license. An L.Ac is an L.Ac. no matter what degree you have. Because what we do in our clinics looks so simple, they think we can’t possibly be doing it right. (This is a subset of “community acupuncture isn’t real acupuncture”.) Surely we need someone with a longer, more expensive degree to diagnose our patients and then tell us where to put the needles.

When acupuncturists bring that up, my immediate response is that it makes no economic sense. A major reason that community acupuncture exists at all is that patients can’t afford conventional market rates. But even treating six people an hour at rates they can afford doesn’t make community acupuncture particularly lucrative, unless your overhead is very very low. By and large, community acupuncturists are not raking in the cash. If you add a supervising Doctor of Acupuncture into the mix, the whole thing falls apart; the labor costs just don’t pencil out. (Not to mention that the math for creating Doctors of Acupuncture doesn’t pencil out, either.)

Ironically, I think the NCCAOM might be suggesting that the diagnostic capacities of a Doctor of Acupuncture could be replaced by an A.I. (but not the labor and accompaniment provided by the lowly acupunks). Having hung around the acupuncture profession for the last thirty years, I can well imagine that the lofty language about A.I. “revolutionizing” acupuncture is referencing some ill-considered scenario like this. The acupuncture profession is just not very good at thinking things through.

I’m here to tell you, though, community acupuncturists don’t need an A.I. or a supervising Doctor of Acupuncture to tell us where to put the needles. Because that’s not how it works! The problems with the idea of A.I. “revolutionizing” acupuncture are the same as the acupuncture profession’s problems generally: aspirations that don’t make economic sense; dreams of progress that aren’t grounded in how most acupuncturists actually practice OR what many patients actually want from acupuncture — and maybe even a lack of respect for acupuncture itself?

1. This happens a lot: somebody in the acupuncture profession says something lofty and I have to ask, But what does that mean? See also: acupuncture education as personal transformation.

2. According to John Pirog, the author of The Practical Application of Meridian Style Acupuncture and The Dark Warrior Guide to Chinese Medicine:

All ancient civilizations attempted to grasp an understanding of the universe by searching for symbols. It is through a study of symbols that we access a part of our consciousness that is able to see connectedness where ordinary logic can see only separation...the ancients believed that such self-evident relationships could be rationally systematized to serve the interests of science. Through the mind’s ability to perceive metaphorically, the wholeness of the universe could become entirely perspicuous and infinite numbers of phenomena understandable through a finite body of universal symbols. And so, we find symbolism at the very core of Chinese medical theory, playing a role so prominent that at times the subject looks more like mythology than medicine. (Introduction, The Dark Warrior Guide to Chinese Medicine Vol I)

3. See also Paolo Freire’s Pedagogy of the Oppressed (on the reading list at POCA Tech!)

4. See: “Graduate Programs Face a Federal Reckoning”, Inside Higher Ed, June 26 2025.