How much AI suffering matters does not depend (much) on how we count AI systems by LeonardDung

Leonard DungCon­fi­dence: I am con­fi­dent in the part ex­plain­ing why ques­tions of AI iden­tity are not quite as im­por­tant as one may think. I am some­what less con­fi­dent in the brief thoughts about how to count/​in­di­vi­d­u­ate con­stituents of AI suffer­ing—the philo­soph­i­cal is­sues there are com­plex and lit­tle ex­plored. I am un­sure how com­mon the view I ob­ject to re­ally is.The cen­tral case for think­ing that the risk of cre­at­ing AI sys­tems that can suffer (e.g. are in st...