I think it does sort of carry that implication.
Thing is, we're of a species which mostly has two subtypes (male and female). So...
1. There's pretty good evidence that human brains are at least partially pre-wired, in general, to subdivide the species that way.
2. And to identify the self as one of these subtypes.
2. And to find members of one of these subtypes attractive, but not the other.
Except that of course, not all of these things always happen, but even when one of them doesn't quite turn out to be true, the others tend to be still there as default assumptions. So even though someone may not show any preference in partners based on sex, gender, or anything like that, the easiest way for most people to think of this is "attracted to both men and women".
I think this is a case where there is probably some benefit to occasionally pointing out in passing that, hey, there's other options here, but trying to replace the word is probably unrewarding; sometimes we end up with words with etymologies rooted in beliefs which have been long-discredited. (e.g., "hysterical".) No real point in trying to fix all of them, I suspect.
So... Yes, the term sort of implies only two genders, but to be fair, at least a fair number of bisexual people are only attracted to people of those genders, and find people who are of other genders less attractive, so it's probably even accurate sometimes. I know at least one person who identifies as "pansexual" to make this distinction.
(Disclaimer: I have no clue at all how I would be categorized, and consider the point moot for the forseeable future, since I'm happily married.)
