Out of curiosity I assume we could have AI divide states into population equal districts, without bias. Id be curious to see if anyone has done that, and how things would change.
This would almost certainly favor Republicans.
It doesn't sound like it would since it is AI and "without bias" but in our current configuration "without bias" favors Republicans. Here is why:
Let's say that Ohio was still a battleground, 50/50 State like it was up until at least 2004. Now just to make the math simpler lets pretend that Ohio had a slightly larger population such that they got 20 congressional districts.
Ok, in theory since Ohio is a 50/50 State and they have 20 districts there *SHOULD* be 10 that lean Republican and 10 that lean Democratic.
That is great in theory but in practice the Democrats are nearly always more condensed. In Ohio you have parts of Cleveland, Columbus, Cincinnati, Toledo, Dayton, and Akron that are basically 80%+ Democratic while the rest of the state is something like 60% Republican.
If you completely ignored partisanship and simply drew geographically compact districts what you would almost always end up with is a relatively small number of overwhelmingly Democratic districts centered on the big cities and a much larger number of less strongly Republican districts everywhere else.