A tweak suggestion that occured to me: possibly if Magic Duel cost a gem to cast (consider it a "focusing cost"), that might even things. Then, not only would the attacker be risking death but there'd be a limit to his/her/its "ammo" supply over time, and resources that could be used for other spells, including other offensive choices, would be consumed.
FWIW, I haven't been running into Magic Duel much, but I realize a lot of people script their mages and that would "force" (ugh, what a biased-sounded word, but I can't think of a better one right now) Magic Duel into the mix if they wish to use it. The spell-chooser AI, however, doesn't seem to select it very often. Mind Burn is the big winner there.
Possibly it is because of the risk factor, but I think the fact there are many spells available that can be sent at Groups of enemies instead of single opponents is more likely the root of that decision. It probably also makes a difference that there are totally safe spells (i.e. Mind Burn) that can be pinpointed onto specific targets.
If you add the cost of a gem to the mix, and the AI will now only use gems if it feels it is really necessary, then that would make computer-guided mages (be they enemy or friendly) all the less likely to cast the spell.
Eh, just my 2¢ there.
I also feel like adding, after my reread, that I've been staying out of this one because I just haven't run into Magic Duel very much. (There's no point in me shooting my mouth off on a topic I just don't know all that much about; I'll save that for the ones I'm more familiar with.
)
I love Astral Magic, and while some of the new spells are very expensive both to research and to cast, they are really, really potent! This, however, would lead to the enemy not wanting to challenge my empowered front-line mages (they need the boost to cast the high-power spells I love) and my Pretender will always have an Astral of 9+ if I go Astral with him/her/it because of the specific blessing effect it gives. I, myself, haven't tried the spell in question more than once or twice by scripting. It just seems like less of a gamble to send a nice Stellar Fires spell raining down on the poor shmucks, but that's just me; experience with the spell (and I do intend to experiment) may well change my feelings on the matter.