Part of this is a problem with the concept of 'passive' in a game with a very limited player-centric perspective.
The AI will go to war. The AI will partake in plots. The AI will engage in feuds. The AI will expand, will fracture, will have civil wars and expend their resources (prestige or gold) needed to continue more wars and use hooks for contracts and so on.
It's just, you know, the AI will mostly do that against other AI, and so you the player will never notice the vast majority of the things the AI does. As a result, the AI can simultaneously be very busy, and very passive, because what they do is not apparent or directly impacting the player.
This is especially true if what the AI does do is run itself into a 'can't start more wars' state- such as tribals using up all their prestige on MAA and thus losing the ability to sustain wars, or feudals going deep into debt. As war is the most obvious AI activity for the player to notice, a lack of war for inability-driven-by-previous decisions is at first glance indistinguishable from passivity-by-choice-despite-ability.
Given that in the early game especially- the 867 start where most players start and spend most of their time- the AI is too economically depressed to afford constant actions, this in turn leads to conflations of passivity (a choice not to act) with inability (the AI is unable to afford to act).
The other part is this- a fundamental misunderstanding of what the Paradox AI is.
Paradox AI is not intelligent in the sense that you understand intellignece. It cannot be coded to be intelligent, because no one knows how to actually do that- there's a reason the most advanced AI in the world are basically language token generators predicting responses off of syntax patterns. There is no concept of the AI 'understanding,' or 'fear,' or 'seeing.' These are not abstractions that Paradox AI can do. This is not 'but they could feel fear if Paradox wanted them to,' this is 'Paradox does not know how to make an AI understand the concept of 'the future,' because the AI does not have the concept of concepts.'
What Paradox AI actually do is follow general build orders (in games with more economic build-systems), sequences of what to build when certain resources are available based on current resource amounts and streams, and make weighted decisions where certain types of decisions (event choices, military customization, war declaration thresholds) are more or less likely based on categories called AI personality traits. When a choice check is made, the code checks for if a sufficient thresholds are met as preconditions (does the character have the necessary requirements to do a war if it wanted to), and then makes choices between decisions on relative weighting (higher personality weights lead to certain decisions more often- but not always- over weaker personality weights). This is true of basically every Paradox game- some balance between AI personality weights and pre-scripted build orders.
In CK3 (and in CK2), these paradigms are known as the
AI personality, which are dominated by
character personality traits. Other traits, including lifestyle mastery traits, can increase AI personality values, but the dominant factors are the AI personality trait overlaps. For example, the AI trait of 'Boldeness' does the following-
Well, the character trait Brave is +200 boldness, and Craven is -200 boldness when the spectrum for decisions is +/- 100. In effect, characters with these traits will sit at the extreme for their entire life, making decisions accordingly.
And CK3 AI personality does drive AI decision making! If you look at how AI with resources to act act over time, it's consistent with what their personality types indicate 'should' happen. This is arguably even more important than CK2 AI personality weights, since the CK2 personality system was far less consistent and liable to change on the RNG of societies and standard event pools. CK3 characters have far more distinct AI personalities for the entire lives, rather than a wash of everyone acting more or less the same due to eventually having more or less the same traits over lifetimes.
But- and this is back to your discontent- weighted decision making is inherently reactive. It is computationally 'cheap'- it's something within Paradox's ability to provide- but it does not make decisions on even the most basic premise of patience. Resources are available, resources are expended according to present conditions and probabilities, and that output is potentially 'characteristic' but has no actual thought behind it.
(Later Edit: This structure is also why the Conqueror trait is so impactful for AI behavior. The conqueror functionally faces no resource constraints to declaring or maintaining wars. The Conqueror isn't 'planning' for future growth- it's just reacting faster since it has fewer forced pauses on expansion actions.)
