A relevant point here seems that while your utility function for humans can be bounded, your meta utility function (evolution) need not be.

I find helpful to push against the epistemics of positive utilitarians, asking whether they would want to spin up instances of their kids suffering rather than themselves.

Especially if the person is high in mental resilience.

A good followup would be what proportion of smart people actually do something useful. And maybe how much luck there is involved in whatever metrics you end up using.whatever