Do you remember Reverse Feeding from Smash 4? With Reverse Feeding, you changed your amiibo’s stats in order to influence its behavior. Having a negative Speed kept them from jumping, having certain bonus effects kept them from using aerials or encouraged them to use special attacks, etc. Basically, Reverse Feeding was the discovery that an amiibo’s stats and effects affected their behavior.
What if that’s still present in Ultimate? Is it possible that we could teach amiibo to perfect shield more effectively by giving them high-Defense Spirits, and then remove those Spirits once it came time for a competition? Could they be taught to use their throws more effectively by adding a Grab stat, or could they be taught to avoid Mario’s fireballs by giving them a Spirit that’s weak to fire?
Bonus effects this time around are highly specific, thanks to their use with Spirits in World of Light. We have Spirits that do more damage to metal enemies, Spirits that increase damage dealt with magic attacks, Spirits that cause water attacks to do more damage and much more. Assuming Spirits did affect amiibo behavior, we could tailor our training regimen to individual amiibo using different Spirits to direct their behavior. (As it turns out, personality types ended up being a purely nominal description of the amiibo, not a toggleable effect on them. Thus we can still write about individual amiibo without having to worry about all of the possible personality types.)
Who knows? Everything is up in the air. As I write this, I’m taking a break from studying for finals. I’ve also finished organically training a level 50 Mario amiibo, and I’ve got a level 1 Mario amiibo on a different training regimen. As soon as I scanned him in, the Learn button was shut off. Now he’s fighting another Mario on a 99 stock match, just to get him leveled up. Theoretically if he’s not going to learn anything then he should be the juiced-up representation of the base amiibo AI… but we should wait to reserve judgement on that until we experiment.
See, if we can take a Mario at level 1 and keep him from learning, then the AI part of his code will stay at its original value, right? What if we take another Mario, put the exact same Mii and name on him, give him the same costume and everything, and have him go through the same process? If they’re both the same except for being physically different amiibo (because both of them will never learn anything) then at level 50, their files should say exactly the same thing. Obviously we’d have to figure out how Ultimate has formatted its files, which Tagmo has yet to do. That being said, we can still view its hex data through Tagmo, and probably many other applications, so we should still have that option open to us.
Anyway, this was more of a stream-of-consciousness piece than anything else. I do believe that these theories are worth being investigated, and once I’m done training this Mario I’ll look into them.