They literally did this in the dev clash- they ensured no AI was present in the game so nobody could see how screwed it would be in 2.2. They then had AI problems when the khan spawned and promptly broke.
- 4
- 2
- 1
Is the AI worse now?
- Thread starter EpicCanadian
- Start date
-
We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
Did..did you miss the section on AI in the changelog of 3.0.0?
- 17
- 9
- 4
- 2
- 1
Definitely still seeing AI's with Food and consumer good shortages (though granted, the main culprit is food now).
Is there anyway of adding an 'emergency measures' type feature with the economy minister, where if the AI has 0 food stock, but 5k minerals and free farming districts (EDIT: which I may add is a frequent occurance when viewing AI's with food shortages), it will exhaust as much of the minerals on farming districts until it's got a +20 surplus or something?
There needs to be some trigger where if an AI has minus production, 0 stock and the capacity to fix the deficit, it should be downing tools in all other departments to fix it (obviously not for all cases and I'm exaggerating to get across the point) as this is primarily where the AI is failing with its economy.
Is there anyway of adding an 'emergency measures' type feature with the economy minister, where if the AI has 0 food stock, but 5k minerals and free farming districts (EDIT: which I may add is a frequent occurance when viewing AI's with food shortages), it will exhaust as much of the minerals on farming districts until it's got a +20 surplus or something?
There needs to be some trigger where if an AI has minus production, 0 stock and the capacity to fix the deficit, it should be downing tools in all other departments to fix it (obviously not for all cases and I'm exaggerating to get across the point) as this is primarily where the AI is failing with its economy.
Last edited:
The 3.0.0 patch notes were 591 lines. The AI section had 11 bullet points; that's about 2% of the total patch notes. if you just want to go by "how is this represented in the changelog?" it's clearly not a priority, hardly even bothered with.
But let's take this further; perhaps the patch notes are not completely representative of how changes affect the players. For that we rely on actual gameplay tests. Did you read the many reports of specific AI problems in this thread, and do you have any comments in response to them?
- 15
- 1
- 1
- 1
Do Paradox consider those 11 lines to be an appropriate amount of improvement to an AI that has been in an awful state for years?
- 6
- 2
- 2
- 1
I'm not going to get into a point by point breakdown, but the number of points does not necessarily how much effort was expended on it.
I've read a number of them but I think it bears mentioning. If you encounter AI issues, discussing them in a general manner here is fine but if you want to ensure that we see and investigate issues you've encountered. Log them on the bug forums, especially if you have saves.
We have a process for getting issues from the bug forum into the bug database. Not for combing the general forum, we need to be efficient with our time.
Lol I wish. Trying to arrange the time of a studio full of people for one MP is difficult enough. We have weekly MP's, but we do most of our testing in single player, both short and long games. Directed and Exploratory.
- 20
- 9
- 3
I can't speak for the company, only myself. But I consider it a step in the right direction.
Nobody is more aware of the issues with the AI than we are, however we are juggling a lot of priorities and resources. You can disagree with the amount of priority AI has been given, and I might even agree with you.
We'll see what the future holds. I'm not at liberty to discuss future plans but there's plenty of discussion ongoing.
- 19
- 12
- 4
- 1
To be fair, Stellaris has so many variables that I imagine it's really difficult for balance to take into consideration them all. Just with different settings on the sliders before you even start the game, a game can be completely different.
Maybe a disclaimer about what the intended settings are, would help a bit.
Maybe a disclaimer about what the intended settings are, would help a bit.
- 1
You know, I've encouraged people to that. I've done it myself, for years. See, for example, this post from years ago which details an AI bug and provides specific reference to source code where the error exists: https://forum.paradoxplaza.com/forum/threads/stellaris-free_amenities-sign-error-2-2-4.1151272/ . It received no replies. I don't actually know if it ever fixed because, as a player, it shouldn't be my job to repeatedly follow up on my own bug report. Maybe it was fixed, who knows.
I suspect that a "challenge" is that AI bugs are too easily de-prioritized. Game crashes? Don't get an achievement you're supposed to get in a new DLC? Clearly identifiable, easy exact test case given a savegame, easy to claim victory once code is fixed.
But... the AI's planets starve, sometimes, for some combinations of governments/techs/ethics? Sure, it may significantly worsen the gameplay experience, but it's not a crash. So maybe it just doesn't get fixed. Perhaps for years.
Over enough years, and enough AI bugs deferred or ignored, you end up with a thread like this. Again and again. Ultimately it will take PDX prioritizing how their dev process deals with AI by PRO-ACTIVELY going after the issue. Can we depend on you all to do that?
I mean, a lot of the problems are so obvious that it shouldn't take dedicated bug reports to be aware of them. Just fast_forward a game on default settings and then go start looking at AI empires and AI fleets. The issues are there to be found.
Here's one with a savegame from 3.0.1 that I reported. https://forum.paradoxplaza.com/forum/threads/stellaris-unbidden-stop-attacking-empires.1468565/ ; it links to a larger general discussion. Many players in reply pointing out similar issues. One player in the general thread going into detail about a fleet AI failure. But ultimately the game is playing basically the same as I observed in 2.6.3. Players bang their heads against the wall patch after patch -- there's only so much reporting and discussion we can do. People create AI mods to fix certain issues. PDX could more aggressively grab fixes from those mods. All the problems are out there, it's a matter of PDX management thinking it's worth spending time on it.
I guess if management has difficulty quantifying how this impacts their revenue stream, they don't appropriately prioritize it.
- 13
- 8
- 1
Glad to heard you.
To be honest I have had the feeling that AI priority has been always very low in this game.
And I can even understand why. Unfortunately all the work made to improve the AI are always very diffucult to sell and priorities and effort in other areas are better rewarded. For me 3.0 is a step forward in the good direction.
Please, give the AI the priority that it deserves. Some of us love the game but we are not happy enough with this AI. I would even buy a "AI improvement DLC" without anything else.
Last edited:
- 4
- 1
One of the things that worked in the AI's advantage before the pop growth changes was it's resource bonuses could be used. Since pops grew fast enough, you could feasibly pivot your priorities quickly. Now, the further you get into the game, the harder it is to pivot without rebuilding infrastructure. Making decisions to correct imbalances is difficult. So even though the AI has tons of resources, it may not have an effective way of using them.
For players at least, it is kind of nice that your output stays pretty steady now. Though digging out of energy/food/mineral deficits can be stressful sometimes.
- 5
Never knew that. Maybe, you should rename it into something else than "Bug Reports". I would never have thought to put AI behaviour problems there.
- 1
One thing I noticed, is that you didn't update the economic plans for the AI. They still aim for the old population numbers. I don't know how much impact those numbers have on the AI, but the early plan aims for 500 pop which cannot be achieved with the new pop growth mechanisms.
Code:
early_default_plan = {
type = early
income = {
energy = 50
minerals = 200
alloys = 100
unity = 50
physics_research = 200
society_research = 200
engineering_research = 200
exotic_gases = 1
volatile_motes = 1
rare_crystals = 1
}
focus = {
energy = 10
minerals = 30
alloys = 50
}
subplan = {
potential = {
country_uses_consumer_goods = yes
country_uses_food = yes
}
income = {
food = 50
consumer_goods = 50
}
focus = {
food = 10
consumer_goods = 20
}
}
subplan = {
potential = {
country_uses_consumer_goods = yes
country_uses_food = no
}
income = {
consumer_goods = 50
}
focus = {
consumer_goods = 20
}
}
# increased food goals for hiveminds
subplan = {
potential = {
country_uses_consumer_goods = no
country_uses_food = yes
}
income = {
food = 50
}
focus = {
food = 10
}
}
pops = 500
empire_size = 1.25
ai_weight = {
weight = 1
}
}- 9
- 6
Have to agree - I actually purchased Nemesis and 2 other DLC's I'd held back on since the lack of improvements with AI and bugs, only to refund them when I realised these problems are still present.
Will very much be looking forward to when I can catch up with the DLC once these very fundamental issues have been resolved.
- 7
highly informative post, describes a specific AI problem related to a new feature, provides source code where the problem occurs, and fixing it would probably make the game better. but, you know, they don't have a "process" for logging reports like this because it's not in the "bug reports" forum. so your post may just be ignored. I guess you also have to write a bug report.
but wait, attachments are now MANDATORY for new bug reports. I guess you could screenshot their own source code, or perhaps provide their own source code to them in the form of a text file.
- 4
- 2
This is actually an excellent pick up, because I thought the new AI planning system was a good step in the right direction when introduced, and adjusting the numbers to suit the new system should be both low handing fruit. and yield better results.
Edit* did you log this in bug reports?
- 5
Serious question for internal PDX expectations - I know as a player I hate seeing red numbers or especially total deficits in the Empire - how much at certain growth points is a growing AI Empire to be undergoing shortages? I saw a lot tabbing around the console.
I did re-rig the population which did affect everyone, and a century later the remaining AIs are all in the green, but sheer tech could've helped as well.
I did re-rig the population which did affect everyone, and a century later the remaining AIs are all in the green, but sheer tech could've helped as well.
- 1
AI is better now. It used to be after 2350 AI literally just stops making buildings, but it's not the case anymore.
- 3
haha, man. that pretty much says it all. for a game released in 2016, the bar should be a little higher than "it now makes buildings after 2350"
- 4
- 1