i can agree that there is a legitimate difference between lawful and unlawful violence in the predictability of their outcomes. however i can’t say i’m convinced that it addresses or even acknowledges the deeper concern.
both systems ultimately rely on coercion compliance through the threat of punishment—actors comply because of expected or perceived future punishment. my concern is that coercive systems don’t scale so well to something global, existential, and economically incentivised like ai development. unlike nukes, ai has massive commercial incentives, and a low barrier to entry; any sufficiently motivated actor with enough compute can defect from any sort of rules set in place.
furthermore, i don’t see anything inherently wrong with debating over this mechanism, but i question the viability of some supranational body ruling over a commercially lucrative product, such as, ai. i mean, companies would have to first voluntarily enter, and only then be bound. if enforcement relies on voluntary participants and incomplete coordination the system remains fragile.
fundamentally i’d like to see a shift away from ‘lawful vs. unlawful’ debates and towards a ‘coercion vs. persuasion’ one. avoiding punishment is instrinsically different from pursuing a shared reward. if anything i’d like to see a stronger argument that coercion—regardless of the lawfulness of it—is sufficient to align individuals, governments, and the actual corporations profiting [or profiteering] who have every financial reason to route around it.
i can agree that there is a legitimate difference between lawful and unlawful violence in the predictability of their outcomes. however i can’t say i’m convinced that it addresses or even acknowledges the deeper concern.
both systems ultimately rely on coercion compliance through the threat of punishment—actors comply because of expected or perceived future punishment. my concern is that coercive systems don’t scale so well to something global, existential, and economically incentivised like ai development. unlike nukes, ai has massive commercial incentives, and a low barrier to entry; any sufficiently motivated actor with enough compute can defect from any sort of rules set in place.
furthermore, i don’t see anything inherently wrong with debating over this mechanism, but i question the viability of some supranational body ruling over a commercially lucrative product, such as, ai. i mean, companies would have to first voluntarily enter, and only then be bound. if enforcement relies on voluntary participants and incomplete coordination the system remains fragile.
fundamentally i’d like to see a shift away from ‘lawful vs. unlawful’ debates and towards a ‘coercion vs. persuasion’ one. avoiding punishment is instrinsically different from pursuing a shared reward. if anything i’d like to see a stronger argument that coercion—regardless of the lawfulness of it—is sufficient to align individuals, governments, and the actual corporations profiting [or profiteering] who have every financial reason to route around it.