1) Reduce the effect of empire sprawl on AI empires by 50% (at least empire size from colonies), or scale it based on the difficulty setting. One of the main ways AI empires regularly cripple their earlygame momentum is by capturing tons of star systems and settling a bunch of colonies. Even with minimum habitable worlds, I will regularly see empires with 5 or so colonies (that are essentially producing nothing and inflating empire size to ~200) within the first fifty years. The effect of this is so pronounced that my Life-Seeded, Aquatic, and Shattered Ring Empires (especially if they have the Rooted trait) are consistently ahead on the scoreboard because they don't spam new colonies. The full solution would be to code them to be better in general about deciding when to expand, and perhaps give certain Genocidal civics -empire size effects, but a simple modifier is possible to create and implement in the next patch.
2) Science Building techs (Xenobiological Labs, Material Workshop, Field Dynamic Center) should be prioritized much, much, much higher by AI. The base effect of these buildings is a 33% increase in the effectiveness of each scientist-type, plus probably doubling the number employed, resulting in a (relatively) massive boost in research. I usually go from 30~ to ~70. When your early game techs take 3-5 years to research, every time the AI does not select these techs means delaying their growth by at least that amount of time. You could also remove/consolidate deeply bad level 1 (and maybe 2) techs like Automated Colony Ships, Ground Defense Planning, Zero-G Refineries to increase the odds they take something useful, even resource job upgrades or fleet.
3) Buff Bureaucrat base output. Human players don't often employ priests and related job types because there are so many other ways to generate unity that it's kind of a last resort. They also don't have nearly as extensive a tech chain to buff their output like Researchers do. Not only would this make spiritualist builds better, it would also help the AI get through Tradition trees, as they always helplessly employ Bureaucrats. A simple +1 would suffice.
And last, the hard option. Allow players to optionally customize each AI empire with a list of 3rd Civics, Traditions, Ascensions, Traits, and Techs to prioritize. Weighted, not guaranteed.
The devs will probably never make an AI that adapts to Stellaris's many systems as well as a human can. The nature of their DLC model means that the AI will fall out of date from time to time, and they will have to work after every major patch to make it functional again. However, even when the AI is fully up-to-date (like in say, 3.14, where some AI empires could hit some impressive fleet and economy numbers) if you look at AI economies and builds closely, they're mostly relying on their outputs being force-multiplied by the difficulty modifier. Essentially, Stellaris AI cannot optimize the available late-game, 100+ year build decisions at all, never has, and it kinda randomly selects whatever.
Expecting Stellaris's AI to make an informed decision about which ascension is correct, or which techs to prioritize, for all the potential builds in this game is silly. But we can expect players to do that. Letting players select a priority list of 3rd Civics, Traditions, Ascensions, Traits, and Techs bypasses the need to create the most sophisticated AI the game industry has ever seen, and allows the AI to scale organically with the playerbase's understanding of the game. Better yet, most players who have criticisms about the AI are veteran players, who are exactly the people most equipped to solve the problem themselves. It turns the question from, 'Hey, why is the AI so crap?' into, 'Well, did you build your AI empires correctly for a challenge?'. All the devs would have to do is make sure that AI empires have a smooth earlygame ramp up without randomly setting themselves on fire.
2) Science Building techs (Xenobiological Labs, Material Workshop, Field Dynamic Center) should be prioritized much, much, much higher by AI. The base effect of these buildings is a 33% increase in the effectiveness of each scientist-type, plus probably doubling the number employed, resulting in a (relatively) massive boost in research. I usually go from 30~ to ~70. When your early game techs take 3-5 years to research, every time the AI does not select these techs means delaying their growth by at least that amount of time. You could also remove/consolidate deeply bad level 1 (and maybe 2) techs like Automated Colony Ships, Ground Defense Planning, Zero-G Refineries to increase the odds they take something useful, even resource job upgrades or fleet.
3) Buff Bureaucrat base output. Human players don't often employ priests and related job types because there are so many other ways to generate unity that it's kind of a last resort. They also don't have nearly as extensive a tech chain to buff their output like Researchers do. Not only would this make spiritualist builds better, it would also help the AI get through Tradition trees, as they always helplessly employ Bureaucrats. A simple +1 would suffice.
And last, the hard option. Allow players to optionally customize each AI empire with a list of 3rd Civics, Traditions, Ascensions, Traits, and Techs to prioritize. Weighted, not guaranteed.
The devs will probably never make an AI that adapts to Stellaris's many systems as well as a human can. The nature of their DLC model means that the AI will fall out of date from time to time, and they will have to work after every major patch to make it functional again. However, even when the AI is fully up-to-date (like in say, 3.14, where some AI empires could hit some impressive fleet and economy numbers) if you look at AI economies and builds closely, they're mostly relying on their outputs being force-multiplied by the difficulty modifier. Essentially, Stellaris AI cannot optimize the available late-game, 100+ year build decisions at all, never has, and it kinda randomly selects whatever.
Expecting Stellaris's AI to make an informed decision about which ascension is correct, or which techs to prioritize, for all the potential builds in this game is silly. But we can expect players to do that. Letting players select a priority list of 3rd Civics, Traditions, Ascensions, Traits, and Techs bypasses the need to create the most sophisticated AI the game industry has ever seen, and allows the AI to scale organically with the playerbase's understanding of the game. Better yet, most players who have criticisms about the AI are veteran players, who are exactly the people most equipped to solve the problem themselves. It turns the question from, 'Hey, why is the AI so crap?' into, 'Well, did you build your AI empires correctly for a challenge?'. All the devs would have to do is make sure that AI empires have a smooth earlygame ramp up without randomly setting themselves on fire.
- 1
- 1
