Yeah, my understanding is that the AI bases decisions about approaching/retreating on relative flux levels. If a ship has low flux while the enemy has high flux, it will play very aggressively, so a falcon that has a flux advantage will chase down its opponent. But if you have a range disadvantage, then you will take damage on shields while approaching before you can deal any damage, which sets you behind in the flux battle, and if that flux offset is too large, the AI decides it's losing and needs to back off and never actually gets into range. This is why capacity can be important for 1v1s, because it means that you need to take more damage to hit that threshold where the AI backs off, so it increases the chances of the AI committing to a fight.
A human understands that you need to commit to the fight to have a chance of achieving anything, but in a fleet context where committing can expose you to a lot of other threats, that can be very dangerous, and so the AI is generally set to play it safe and not commit unless it has a flux advantage unless you give it specific orders. I don't think that's totally unreasonable, but it definitely does not work as well in 1v1s. Eliminate orders can fix that to some extent though.
The aggressiveness of your AI does have a big impact too since it determines the tolerance of the AI to being at a disadvantage. I think this is the main tool the player has to work around these mechanics. Aurora and Fury basically require agressive AI IMO. Steady is too timid, and Aurora/fury don't have the hull/armor to tolerate reckless IMO.
I also feel like the AI doesn't really use mobility systems to achieve specific goals (like getting into range or chasing something down) with much consistency. It feels like it mostly just pops it randomly and doesn't do much, although I'm sure it's doing more than that under the hood. It feels like scarabs in particular love to pop their system and then just hover out of range.