Source: Brad Hooker explains how the desires of our empirical human nature can never be the origin of moral actions, according to Kant.  Self-love distorts all our natural chocies, and univeralisability is closely linked to benevolence or concern for others.

Kant’s ethics is grounded in the distinction between hypothetical imperatives and categorical ones.

By ‘hypothetical imperatives’, he means imperatives that tell you what you must do in order to get or do something you happen to desire: for example, ‘if you want a good sleep, don’t drink a gallon of caffeine at bedtime’, or ‘if you want to be trusted, always keep your word and tell the truth’.

‘Categorical imperatives’, on the other hand, tell us what to do regardless of our desires. I’m required to tell the truth even if I don’t happen to want to. The same is true of my other moral duties.

But what could bring about intentional human action except desires, preferences, inclinations, and the like? Kant thought that unless there is something that can supply an answer, morality is a sham. He thought we aren’t responsible for our desires and preferences. (We couldn’t help the dispositions we were born with, nor the training we received in our formative years.) And if these desires, preferences, etc., completely determine our behaviour, then we can’t really be held responsible for that behaviour either.

Kant thought that our desires and our beliefs about how to satisfy them are not the only things that could guide our intentional behaviour. We could act from duty. And what determines what duty requires? Kant’s answer is: ‘Since I have robbed the will of every inducement that might arise for it as a consequence of obeying any particular law, nothing is left but the conformity of actions to universal law as such, that this
alone must serve the will as its principle.’