Talk:Confusion

From bg3.wiki
Latest comment: 23 September by Quaris in topic Table for chances of different effects
Jump to navigation Jump to search

Table for chances of different effects[edit section]

This page could really do with adding the table for the chance of skipping vs madness AI taking place vs other effects. I had a little look for a source that would list the probabilities, but couldn't find one. Might be easiest to look in the game files. Galois (talk) 02:43, 23 September 2026 (CEST)Reply

That's a good question. There really are only two outcomes each with a 50% chance. If the target rolls a 10 or lower on a hidden, unmodifiable d20 roll, it gets Confused Stupor which stuns it for a turn. If it rolls an 11 or higher, the target takes its turn, but with its decision making driven by the madness AI archetype.
What the AI decides to do on its turn is hard to say. The madness AI archetype is purposefully lobotomised to be as dumb as possible. Essentially, it removes all factors like "try not to hit yourself with AoE", "prioritise attacking wounded enemies", "deprioritise attacking summons", "deprioritise very high AC targets", "avoid provoking opportunity attacks", "avoid harmful terrain", etc. from the decision making. It also prevents (or heavily deprioritises) using any control actions like Hold Person in favour of simple attacks.
In my experience, this means the AI randomly selects a target (friend or foe) to attack each turn. I have never seen the AI choose to "wander around aimlessly" like the spell description alludes to. NtCarlson (talk) 04:20, 23 September 2026 (CEST)Reply
In my opinion, describing what happens to afflicted targets is important enough to put somewhere in the main section of the condition (or maybe spell), instead of burying it in the notes. Though the more detailed stuff about what actions the AI prioritizes should probably stay in notes.
(Since that seems potentially controversial, I won't make such an edit yet.)
Thanks for adding this! Quaris (talk) 05:25, 23 September 2026 (CEST)Reply