Well, the AI vs. human racers thing is the genesis behind the whole idea. As I see it, the main advantage of the AIs is their precision timing and fine control, but they're not necessarily any smarter, y'know? So if you introduce something where the humans have a way around such things as acceleration tolerances (depending), timing, cutting things closely... it comes to a contest of who can plot out the cleverest path though the obstacles.
So maybe not breaking things between humans and AIs up at all. If you, a human, can come up with a support AI that can help you do things more precisely, than good... or if an AI comes up with new software, also good. But one way or another it comes to how well you can predict how the course is going to be as it relates to where everything is throughout the race.
I mean, take a chaotic asteroid field (of sufficient density that things are interesting; yes things are more spread out than in movies, but velocities are higher too). Movement of rocks can be predicted with software to a certain good degree... but what about changes in velocity vectors due to collisions, added fragments thrown up from them, disturbances of the rocks from exhaust backwash (or whatever) and innumerable other factors? One way or another, the pilots have to figure out the most efficient way through all of it and... well, let's just take it as a given that it can't be completely accurately modeled.
For an example why I'm thinking AIs wouldn't just automatically dominate because of reflexes and whatnot... say you have two ships, one with an experienced and clever human with something akin to the support software I described in the OP, one with an AI who wants to just brute force it, both in otherwise identical ships. So the AI is going to go in a straight line and run collision avoidance; whenever anything big is in the path, it skims the surface of the obstacle at the last moment. The human, on the other hand, makes educated guesses about where there's going to be gaps in the rocks based on prior experiences, and plots his vector through those. Because there's less hard manuvering for evasion, speed stays up and the course takes less time.
Now of course the AIs can try to model things to do guesses, the human pilots can add software to do the same sort of emergency collision avoidance that the AIs in the brute force example was doing too, but at this point it's again getting to the "which individual human or AI is cleverer at knowing how things are going to move and finding a way through it?" thing. "Cleverer" in this instance including a combination of vehicle design, software design, and ability to figure paths.
At least that's how I'm seeing it.
So maybe not breaking things between humans and AIs up at all. If you, a human, can come up with a support AI that can help you do things more precisely, than good... or if an AI comes up with new software, also good. But one way or another it comes to how well you can predict how the course is going to be as it relates to where everything is throughout the race.
I mean, take a chaotic asteroid field (of sufficient density that things are interesting; yes things are more spread out than in movies, but velocities are higher too). Movement of rocks can be predicted with software to a certain good degree... but what about changes in velocity vectors due to collisions, added fragments thrown up from them, disturbances of the rocks from exhaust backwash (or whatever) and innumerable other factors? One way or another, the pilots have to figure out the most efficient way through all of it and... well, let's just take it as a given that it can't be completely accurately modeled.
For an example why I'm thinking AIs wouldn't just automatically dominate because of reflexes and whatnot... say you have two ships, one with an experienced and clever human with something akin to the support software I described in the OP, one with an AI who wants to just brute force it, both in otherwise identical ships. So the AI is going to go in a straight line and run collision avoidance; whenever anything big is in the path, it skims the surface of the obstacle at the last moment. The human, on the other hand, makes educated guesses about where there's going to be gaps in the rocks based on prior experiences, and plots his vector through those. Because there's less hard manuvering for evasion, speed stays up and the course takes less time.
Now of course the AIs can try to model things to do guesses, the human pilots can add software to do the same sort of emergency collision avoidance that the AIs in the brute force example was doing too, but at this point it's again getting to the "which individual human or AI is cleverer at knowing how things are going to move and finding a way through it?" thing. "Cleverer" in this instance including a combination of vehicle design, software design, and ability to figure paths.
At least that's how I'm seeing it.