I don’t get it.
Goodhart’s law is the tendency, when using a proxy to measure a desideratum, to optimize for the proxy, rather than the desideratum.
In this case, the desideratum is feeling genuine regret about harming others, and the proxy is apologizing.
I don’t get it.
Goodhart’s law is the tendency, when using a proxy to measure a desideratum, to optimize for the proxy, rather than the desideratum.
In this case, the desideratum is feeling genuine regret about harming others, and the proxy is apologizing.