My point of view is that in humans and animals in general, emotions are largely a chemical response in the brain. We might not fully understand how those processes interact, but we do know that certain chemicals cause certain feelings, and that there is a mechanism in the brain governing emotion that is notionally separate from our ability for rational thought.

I am willing to concede that it might be possible for a sufficiently complex computer to accidentally or in a way not entirely within our understanding to develop the capacity for rational thought in a way that we would recognise as sapient, or at least animal level intelligence.

I am not willing to concede that such a computer could develop a capacity for what we recognise as emotion without it being intentionally designed in, and if it's designed we necessarily need to understand it. This happens in fiction a lot because it's more compelling to anthropomorphize AI characters, not because it's particularly plausible.