Beginner’s Mind is a special project collecting insightful essays written by US college students who have attended experiential-learning courses related to Buddhism. Some of the authors identify as Buddhists, for others it is their first encounter with the Buddhadharma. All are sharing reflections and impressions on what they’ve learned, how it has impacted their lives, and how they might continue to engage with the teaching.
Carter Kawaguchi wrote this essay for his Spring 2026 course on Buddhish Economics at the University of Southern California (USC). Carter is a senior studying Computational Linguistics, Business, and Design. He is interested in product management and design and bridging language, technology, creativity, and community.
Cultivating Collective Cetanā through AI De-biasing
For the topic of my essay, I have opted to focus on a specific, leading force reshaping the modern economy: artificial intelligence (AI). AI’s prevalence in society is now irreversible, and we will only continue to see massive economic impacts from this development.
However, de-biasing in AI stands as an increasingly relevant practice. Many philosophical and moral concerns have arisen as these LLMs (large language models) imply massive consequences to the perception of truth and the dissemination of knowledge, and are being utilized across almost all industries. De-biasing offers a unique opportunity to apply Buddhist processual philosophy to this now mainstream tool.
Prevailing perspectives on bias view it as a technical defect rather than a generator of precarity and controller of perceived truth. I propose that this practice of de-biasing in AI can be reinterpreted as the cultivation of collective cetanā (Skt. volition). By approaching de-biasing in AI with a processual philosophical mindset, we can help establish collective cetanā from iterative, intersubjective reviews from diverse stakeholders to reduce precarity and establish a society that functions on a more united and interconnected state to generate change.
AI can be found in nearly every industry and space in our modern economy and society, making it central to creating an economy of care and low precarity. De-biasing AI is critical to this process as these algorithmic systems are increasingly at the center of mediating access to work, resources, recognition, and knowledge. We can see this directly with the rise of companies whose sole purpose is to apply AI to existing processes/industries, such as credit scoring, hiring, and welfare eligibility.
The knowledge that these models are dispersing have real-world implications for many millions of people around the world—potentially unfairly increasing suffering and duhkha for individuals. With the Cakkavattisutta (DN 26) framing poverty and material deprivation as root causes of crime, violence, and moral decline, it is clear how errors in AI systems can rapidly expand in impact.
AI de-biasing is easier to picture as collective cetanā when analyzed alongside Honeybee Democracy and Pyrrhonian-Buddhist skepticism by illustrating how intention and judgment emerge from distributed processes rather than a single mind.
In HoneybeeDemocracy (Seeley 2010), we observe that no individual bee “intends” to control a new nest site by themselves. It is a collective action where the scouts explore potential sites, express their opinions of quality through their waggle dance, and the hive gradually builds quorum for the best site. The hive helps us reframe intention not as something within any individual but as the way that a network leans toward a certain outcome based on collective feedback and experience. AI should be modeled similarly where no data, perspective, and so on, holds precedence over another, and the users, domain experts, engineers, and regulators all play the role of a scout, presenting their local knowledge.
Continuing, in Jonathan Gold’s analysis of Pyrrho’s Buddhism, we can derive a new perspective on the outputs of these models (Gold 2023). In the text, we observe counsel being informed not by fixed views but by testing ways of living and experiencing their consequences. In AI, we can apply this perspective by treating model outputs as provisional, being transparent about uncertainty, and building institutions around revising models, metrics, and policies to minimize bias and harm. Together, these two examples help us better understand collective cetanā in AI de-biasing as structural distribution of power and influence across the social and technological networks of AI, and cultivate a shared practice of epistemic humility and constant revision.
Now that we understand what it means to view AI de-biasing as an act of collective cetanā, we will present three strategies that can be applied now to build a more interdependent economy. These strategies transform AI from an isolated, technical artifact into an integrated social and technical process deliberately oriented toward reducing precarity and fostering shared welfare.
First, widespread socio-technical auditing should be employed by AI companies across every stage of their product development process. From data collection, feature design, training targets, deployment context, and UI/UX, AI companies’ models and products should be designed around reducing bias. Applying an auditing process at each stage would help reduce bias in two ways: exposing the model/product to diverse viewpoints and ensuring consistent processual analysis at each stage.
Second, welfare-centered fairness metrics should be encoded to ensure a commitment to reducing precarity. Existing metrics are focused on improving technical, profit-centric areas such as speed, tone, churn, and so on. Shifting these metrics to those that are not only optimizing accuracy or parity but also for social welfare would serve as a statement of institutional intention.
Third, ongoing participatory governance that include all stakeholders and users of these AI tools will help to quality check model outputs and monitor and revise high-impact AI features. These standing participatory bodies of laypeople, experts, and officials should have a space to deliberate on the AI models with access to verified information and peaceful facilitation. This, in practice institutionalizes a “sangha” of sorts to establish a community of practice that collectively revises AI tools.
It goes without saying that AI debiasing will serve an increasingly important role in our society and economy. This process, as we have seen, should not simply be understood as a technical fix but as a practice of cultivating collective cetanā. It is critical that all individuals that are affected by AI tools are given intention in the models. Developers, institutions, users, and policymakers must have their knowledge considered in a collective decision-making process for AI.
Progress can be made toward a collective cetanā-focused AI industry through regulation as well as institutional intent to improve welfare and reducing precarity over profit and efficiency. The central question for AI in our economy and our world is no longer simply what AI can do, but how can we ensure these tools remain a collective action to improve the lives of all.
Reflection on AI use in this essay
The process of writing an essay with explicit use of AI was refreshing. Especially for the various terms that I struggled to grasp in this course, AI assisted in helping refresh my knowledge and apply these terms in the paper. My approach to using AI was very structured and most of my prompts were extensive. Each of my prompts are rooted from my original thoughts and intentions and used AI to assist in structuring and providing alternative perspectives as I was writing.
Especially as I did not want to misunderstand any of the texts or concepts we were using, AI assistance was helpful. As my essay was centered around AI and its interplay with bias and knowledge, it was certainly interesting to think about as I was writing the essay. I was essentially asking AI how it could fix itself. Its capability to show this minority perspective on how the industry should progress is slightly reassuring, however, I would love to see the policies outlined in the essay be actually implemented as I believe actions like those would help create more equitable and socially-good technology.
References
Seeley, Thomas D. 2010. Honeybee Democracy. Princeton, NJ: Princeton University Press.
Gold, Jonathan. “Pyrrho’s Buddha on Duḥkha and the Liberation from Views.” Journal of the American Academy of Religion 91, no. 3 (September 2023). https://academic.oup.com/jaar/article/91/3/655/7606269
Related features from BDG
The Timeless Art of Being: A Contemplative Resistance in a Burnout Society
Finding Stillness in the Storm
No-Self and Social Suffering
