Some thoughts on "AI" and Yoga
I have shared about this in various ways here before, but in preparation for an upcoming Yoga Alliance focus group on AI and Yoga, I decided to organize some of my thoughts a bit on the topic and simply share this here for discussion, etc. This is not exhaustive and simply a starting point to expand from.
Some folks here in the past have asked "how exactly is AI antithetical to yoga philosophy / yoga?"
Here are some less than complete thoughts on the incompatibility of "AI" and yoga, framed inside of many core principles from yoga or yoga philosophy. I have combined some similar principles together in some cases. These are obviously just my own interpretations and I have not cited anything in this post - no one pays me a penny to write a cited paper (or un-cited post) on Reddit... If you would like sources / citations / links for anything, feel free to ask in the comments and I will do my best.
To me, so many of these points seem so painfully obvious that I almost didn't write some of it out for fear of "well, duh...", but I tried to be as general as possible for the sake of presenting in the information. There is a lot more expansion I could add to many of these points and I feel certain that I have missed some things, as there's always more to say. This is not normally the format (bullet points, etc) that I share in on Reddit, so please feel free to add your own and/or add to the discussion in any way you see fit.
*footnote - Many use the term "AI", but what we are actually talking about (typically) are LLMs (large language models), the tech industry that develops and sells them and those that fund them.
ahimsa (non-violence) - Highly concerning environmental impact / damage via data centers - greatly increased power & water consumption / demand on an already heavily leveraged planet, air pollution, noise pollution. There are also so, so many instances coming to light now (many of them being litigated in the form of major lawsuits) of AI/LLM-induced mass murder, suicide, psychosis / delusion and otherwise general psychological damage to society at large via methods of model sycophancy, addictive tactics of keeping users engaged with the tech and isolation / erosion of sense of reality and social connection to the actual world and other people. *A note should also be added that the major AI / tech companies have used human labor in developing / economically under developed countries to "train" their models in grueling and psychologically damaging methods that include repeatedly exposing humans to CSAM and other heavily explicit material in the process of model "training".
asteya (non-stealing) - likely the largest single IP (intellectual property) theft of the modern age. There is no getting around the fact that LLMs are trained on stolen data that essentially none of us consented to nor were compensated for and this continues to be the case. I personally think this is a foundational and crucial point for understanding the complete lack of ethics / values on display in the tech industry; however, many seem to gloss over it. There is also the point of the tremendous and insatiable appetite for greed and power from the tech industry and their financial backers. Resources of a massive scale that could be going towards much more helpful and useful things have been and are still being funneled continuously into this tech industry with hopes of collecting on their return when the times comes. The time, energy and psychological resources of society are being involuntarily stolen for the potential profit and control of the tech industry.
satya (truth), viveka (discernment), asato ma sadgamaya (lead me from the unreal to the real) - There is no concept or sense of truth or discernment in an LLM; it is simply a mathematical equation that is based on predicting what the statistically likely next character or word is (based on its training data). The training data could literally be anything and has no amount of discernment or context attached to it. LLMs do not think, they calculate based on training data...that's it. This is why you have the hilarious example from a recent post where someone said they asked an LLM "why do yoga teachers teach with their eyes closed?" only to receive an answer that stated that it was to encourage pratyahara...as if that was an acceptable and correct answer. Anyone who as ever taught even a single yoga class knows that this is incorrect and completely out of context for 98% of any asana class, but this was what apparently was most likely statistically based on its training data. It has no reliable or skilled way of interpreting which data is correct, relevant, in proper context, and ultimately useful, particularly for an experiential and embodied practice such as yoga.
Embodiment - AI / LLMs have no body. Period. There is zero frame of reference or actual context that an LLM can draw upon that is its own and isn't simply something that it scraped as "relevant" based on its training data, which is highly variable in quality. They do not move, breathe, feel or think. They are complex and extremely expensive calculators. There is no good reason to ask a calculator about embodiment, especially not at what this truly costs.
jnana (knowledge), karma (action without attachment to results / skill in action) - Continuing on the above, what does this say about the state of knowledge and what it means to truly gain / hold it if it can be conflated so easily with what these theft calculators confidently provide? Where is the sense of true learning, effort for the sake of effort without attachment to the results, learning from our own mistakes and correcting in ways that you can only do through experience and your own personal knowledge? Why are we so quick and eager to outsource this for the facade of convenience and efficiency?
brahmacharya (appropriate use of energy), aparigraha (non-grasping) maya (illusion) - this can be tied closely with the points in ahimsa regarding actual planetary energy use, but I would argue as well that so much of the individual and collective energy of society is being spent / wasted in frustration, anxiety, rage or worse with the ever present "growth" and proliferation and endless propaganda of AI into every facet of our lives. Psychologists have already termed "AI anxiety" as something prevalent enough to create a name for it in less than 3 years since public unveiling of Chat GPT. I am not a litigious person, but I think society as a whole has a case for "emotional & psychological damages" against big tech simply for this alone. The toll this takes on us daily should not be minimized and it feels like a direct opposition of living with the concept of brahmacharya. What sense of brahmacharya or aparigraha is there in using a technology that actively encourages you to use it more and more and more and is subsidized (by VC investors) to be so accessible, available and ubiquitous. Where is the sense of healthy moderation, lack of greed and taking only what we need from this AI-crazed approach? What type of illusion is being presented at large to society in the form of AI, when the reality is actually much different?
svadhyaya (study), sadhana (practice) - This feels pretty obvious, but over-reliance on a machine to "think" for us and "give us answers" actively erodes our own willingness and strength / capacity to do that for ourselves. Why would we need to earnestly study something if we can just ask the chat bot to tell us the answer? Why would we need to have our own steady and thoughtful practice if we can just download the AI sequencing app that plans our classes for us? Why should we reflect on our own experiences, struggles, studies and insights if we can just ask Chat GPT for "inspiration for class themes"? What is our role as a teacher and what are we actually "teaching" if we outsource core elements of the experience and work to a calculator?
atmavichara (self inquiry) - similar to the above, but why should we do the difficult work of self inquiry when we can ask the chat bot for the bastardized cliff notes? One of the primary methods of yoga is self inquiry with the goal of knowing yourself well enough to be able to make the right decisions in moments of both simplicity and complexity. Yoga could be thought of as the active process of self inquiry and searching for truth via discernment, clarity and embodiment, among others things. Why would we want to bypass this beautiful process of yoga simply to receive a stolen calculator answer that is mostly likely not even fully correct?
Do we really believe in these principles and want this to be represented / supported in the way we practice and teach yoga? I have a really hard time seeing the tech industry and AI / LLMs being truly inline with so much of yogic philosophy and living.
As a preemptive response to those who are pro-AI and think it is a net positive to society - do some research on the environmental impacts, the philosophical framework of the movement, psychological damages to society and the financial structure / (insanity) of the current tech industry and ask yourself: "is what we have in the form of "AI" actually worth what it is truly costing us?"
Feel free to add your own points to the conversation, as I am just one person and know I left some stuff out.