Sophecs
Menu
“Treating others like oneself, neither kill nor incite to kill.”

Dhammapada, verses 129 to 130 (trans. Bhikkhu Sujato)

Buddhist Ethics

The passages Sophecs uses are Pali, from the Theravada canon, and passed through long oral and editorial transmission before being written down, so they are witnesses to an early tradition rather than transcripts. Damien Keown reads it as a virtue ethics organised around nirvana; Charles Goodman as a consequentialism of character. Sophecs follows Charles Hallisey and Jay Garfield in declining the choice, taking the tradition on its own terms: non-harm, intention, compassion and liberation. Intention is what the Western categories keep mislaying: an act is weighed by the volition behind it as much as by what follows.

The discourse to Rahula gives the method. Before acting, while acting, and after, you reflect on whether it tends toward your own harm, another's, or both, and if it does you stop and say so. That is an impact assessment with no end date, an awkward and useful thing to hold a deployed model against. The question is not whether the launch review passed, but what the thing is doing now and who it is still hurting.

What Buddhist Ethics gets wrong

Damien Keown points out that Buddhist traditions carry a great deal of moral teaching and little systematic moral philosophy of the kind the West built, the emphasis having fallen on liberation and monastic life rather than politics. For AI governance that is a real gap. Compassion and non-harm do not tell you who owns a dataset, who audits a model, or how somebody appeals a decision it made about them. Charles Hallisey answers that the missing theory may be context-sensitive reasoning rather than a failure to theorise, which makes this a reconstruction problem, not a verdict. Utilitarians would press anyway: a school weighing intention this heavily lets a well-meaning system off for damage a careless one would be condemned for.

Take the quizArgue a motion from it

The other five