The error you're making in trading off a bit more discomfort for an already-marginalized group against any discomfort for the mainstream is exactly the reason having someone in an AI ethics board who doesn't see being a trans person as valid was a problem.

It's also the reason we need more ethics boards, because it's a common error technical training leads a person to, where we're taught how to optimize for common cases and defend against unusual cases---when the things we use aren't people and don't care if they're considered "common" or "unusual."