Marc van Meel

Marc van Meel

Hi I’m Marc!

I speak and write about the intersections of technology, society and philosophy.


What Buddhism Cannot Do For AI Ethics

Share on twitter
Share on linkedin
Share on facebook
Share on whatsapp

Do the fields of Artificial Intelligence (AI) and AI ethics really benefit from more diversity, in particular Buddhism? AI systems are becoming increasingly more autonomous and impactful, forcing us to make our implicit ethical decisions explicit in their source code. As a result, we keep being confronted with the lack of a universal approach to ethical decision-making with respect to AI. As if aligning AI with our core values wasn’t difficult enough, we still even haven’t figured out what these shared values are. This is known in AI as the Value Alignment Problem.

To overcome this problem, by means of diversification, in AI ethics Buddhism is often brought up [1] [2]. But where does the appeal of Buddhism come from? And what can Buddhism really contribute to the ethical assessment of AI systems? I will argue why the Western attraction to Buddhism is not without reason, and why Buddhism won’t just solve the ethical challenges presented by AI.

The New Crusade: Western Ethics

The predominant view in AI ethics conferences today is that AI, and technological artifacts in general, are social constructs. This means that technologies are not neutral, but are embedded in the social relationships of the society in which they are developed. Technologies incorporate the power dynamics of our society as a means of control. This confluence of technology and capitalism gives rise to the fear of technocracy; a society in which technologies are instruments of dominion. This is the Frankfurt School account of technology and is not the subject of this blog, but it helps you to better understand the origin of the argument that is being put forward today:

  • Premise: The West produces most of the AI ethics documents (norms, guidelines, etc.), affecting people and societies all around the world.
  • Premise: These documents primarily reflect our Western value systems and exclude the perspectives of non-Western societies.
  • Conclusion: Through this hegemony of AI ethics, the West has imposed its own ethical narrative on the world, controlling behavior.

western crusade ai ethics

The claim that documents with respect to ethical decision-making in AI are dominated by Western ethics is justified in my estimation. Documents containing AI policies, guidelines, norms, principles, etc. almost exclusively exist out of the combination of deontology, to capture the motivations and responsibilities of Data Scientists, and consequentialism, to capture the outcomes for the affected stakeholders.

The solution offered to this problem is straightforward: all intellectual traditions around the world and different approaches to ethical decision-making, should be included. This is the point where attention is drawn towards Eastern intellectual traditions and the values of Buddhism; compassion, the elimination of suffering, etc. On the surface level this seems pretty reasonable and convincing. Now let’s delve a bit deeper.

We Killed God, and Other Uncomfortable Truths

The Western attraction to Buddhism is not without reason. In fact, we have already imported Westernized elements of Buddhism into our societies. Why? To combat the nihilisms from scientific truth and the abandonment of Christianity.

Nietzsche is often quoted for the part where he prophesizes the dead of God, but he was just as concerned with the fallout from the abandonment of Judeo-Christian values in Europe. He feared the attraction of Europeans to totalitarianism in the wake of nihilism. He therefore argued for a revaluation of all values. Nietzsche was right, as Christian values are being abandoned by the rise of atheism, the need for purpose and meaning remains. This gap is now partially filled by Buddhism, as mindfulness and meditation are practiced to find solitude in our stressful neoliberalist consumerist societies.

God is dead. God remains dead. And we have killed him. How shall we comfort ourselves, the murderers of all murderers? … , what sacred games shall we have to invent?

Friedrich Nietzsche

At the same time we have to contend with uncomfortable implications of scientific realism and naturalism. Consciousness seems to be the exclusive result of information processing in the brain, further confirming that we have no free will. Determinism is the more plausible view, with compatibilist trying to salvage whatever is left of our original notion of free will. We find ourselves in a predicament: science telling us that there is no free will, versus our irreducible subjective experience of being free agents. This is another gap where Buddhism brings comfort, because in the Buddhist tradition the self (anatta), or soul, is an illusion.

With both Christianity and the scientific method, by making truth a value, we have essentially shot ourselves in the foot and we have set the door wide open to nihilism. In our desperate pursuit to cease being objects, free either from God or from nature, we made the horrible existential discovery that we aren’t subjects. But there is a game in town that embraces this emptiness, and that is Buddhism.

Late For Yoga – The Problem of Buddhism

Suffering exists in both Christianity and Buddhism, with one fundamental difference. Where Christianity is centered around the voluntary acceptance of suffering, Buddhism is instead focused on the elimination of suffering (dukkha) by distancing yourself from cravings and desires (tanha). It is for this reason that Buddhism fits capitalism better than Christianity. It offers a way of relief from the stress in our society, makes you a more productive contributor to society by doing so, but leaves the root causes in tact. Hence, the irony in people rushing late for yoga class. Buddhism perfectly augments and doesn’t change neo-liberal capitalist forces. The same forces which voices in the AI ethics community are so often critical of.

buddhism emptiness void accountability

Embracing emptiness does not lead to improved accountability, it leads to less. The most wicked examples of this are found in the practice of Zen Buddhism under Japanese imperialism [3] and the appreciation of Buddhism by Nazi ideologues during World War II. A total decoupling between oneself and substantial reality makes even the most horrendous actions permissible, because all that remains are just experiences, no accountability, no guilt.

But Buddhism seeks to reduce pain and suffering. Yes, but in Buddhism the cycle of life (samsara) is in itself suffering. Enlightenment (nirvana) is only reached if you can let go from this. From an AI Safety perspective, this is not something you would want to teach a powerful AI system. It doesn’t take much for a powerful optimization agent to reach the conclusion that “eliminate all suffering” is best achieved by “kill al humans”.

I don’t mean to single out Buddhism here, on the contrary; the point I’m making is that Buddhism has its flaws and sins, just like all other ideologies. It is therefore naïve to think that Eastern traditions, by default, do a better job than Western ones. We need to understand the strengths and weaknesses of all traditions. Now before you argue that accountability is itself a Western imposed value – would you really want to live in a world devoid of accountability?


I am not a critic of Buddhism, nor of the usefulness of meditation. I am skeptical thought of surface-level solutions pertaining Buddhism, which, in my estimation, originate from 1) a one-dimensional view of technology and 2) a Westernized analysis of Buddhism. Buddhism was originally designed for the individual to reach enlightenment, as described by use of agricultural metaphors, not as a means to solve complex societal issues in AI.

If there is something to be learned from this though, it is that we in AI ethics must first look inwards, instead of outwards. We need to have an open and honest discussion about AI as a technological artifact first, more than just a social construct. We need to acknowledge that there is something inherently dangerous about a technology that generalizes based on mass data and we need to ensure ways to stay in control. Second, we need to open the AI ethics debate to the general public. There is a treasure of intellectual diversity to be found, not exclusive to academia.



Share on twitter
Share on linkedin
Share on facebook
Share on whatsapp

More ideas

Leave a Reply

Your email address will not be published.