Is intelligence inherently kind? Or the opposite, it is malevolent?

Humans are capable of some really fucking awful things, just terrible. Torturing animals and other humans, rape, mass murder, psychological warfare, malicious vengeance, etc.

If a Singularity is triggered, and a Superintelligence comes into existence (as in it can actually think, and better than a human - not simply better at solving tasks) would it be kind, malevolent, or taught to be one way or another?

I assume most humans are evil due to brain abnormality and/or brain damage. A digital superintelligence wouldn't have these problems. But it also wouldn't have evolutionary pressure to be kind to other people.

I suppose an AI's goals truly have to be put directly into the codebase. It can take on any form or personality, correct?