makeinu

I disagree. Suicide is, by definition, an intentional act. However, semantics never wins arguments. So even if you wanted to call it suicide, I would still say that the driver is guilty of a much more egregious moral crime.

The logic works like this:
Statistically speaking, if you do anything enough times you will end up killing a few people. Therefore, defining "wrong" as anything that might kill anyone is a useless definition with no decision making value because it would place all decisions in the same category ("wrong").

A more useful definition would be to define "wrong" as making a decision which incurs a significantly higher probability of killing someone than any alternative decision available to you. I use this definition because it has a number of appealing properties. For example, by this definition, minimizing the number of wrong decisions you make (or the overall wrongness of your decisions if you don't necessarily want to use black and white categories) also minimizes the number of deaths you expect to cause. It can also easily be generalized to include other kinds of wrong and right, such as hurting people, loving people, helping people, etc.

If you have a definition which you think might work better with the kinds of statistical observations that we call "the real world", then I'd love to hear it. But it seems to me that you're trying to apply deductive reasoning to a statistical situation, which can only lead to contradictions.

No, because an activity which could cause death to others may not necessarily be likely to cause death to others. See above about statistical reasoning.