Plans for the AI?

from Sins2 Forums

I am curious as to what the intentions are behind the playstyles, the difficulties & if the AI will be able to lean into the unique traits of the subfactions. I wasn’t sure on how to simplify this down to just a couple of questions, so I opted to flesh out concepts as examples of what I’m inquiring about.

For context, I play large maps in FFA and against AI. I’m not expecting the AI to act like a human or to use meta tactics when it comes to battles (AI is a strong word for what we have access to). I’m more interested in seeing the AI behaviors hold true to the playstyle selected, the difficulty selected & any “overrides” that originate with the faction choice.

1] Playstyles
I’m not sure what the intention is for the playstyles? For Sins2 I don’t know if it makes sense to keep them either. The options are Aggressive, Defensive, Economic & Research – but what do these settings actually mean in terms of what the AI will do? To me, I would assume that these are intended to be the behavior set for the entirety of a game (or until the AI is set to a different mode when a game is loaded). A human player would toggle between these “modes” meanwhile the AI would instead stick to one. Again, I’m not looking for the AI to be like a human player. These are all strategies and every strategy is going to have weaknesses/blind spots – as such the AI should be risking as much as a regular player would by committing to a particular playstyle. So I’m more interested in how these AI ‘types’ will interact with the game's systems.

I’m also assuming that the AI playstyles are intended to be exaggerated, pretty much to a fault. They could be visually recognizable by players (components used, upgrades done, planet tracks upgraded). I would assume something along the following....
-=-=-=-=-=-=-=-=-=-=-=-=-=-

  • Aggressive
    • Fleet – So far as fleet composition it would probably want heavy-hitters to end battles quickly?
    • Components – uses offensive ones
    • Defenses – will protect the homeworld automatically via Starbase & planetary shield. Might also attempt to fortify itself on planets containing artifacts?
    • Research – Will unlock fleet supply and new ships. Will research ability to colonize planets. Otherwise prefers to spend resources on fleet first & will randomly research items once it’s floating some value of resources
    • Planet development – Minimal, as focus is on expanding as much as possible. Unlikely to fully develop non-homeworld planets. Most likely to only put 1-2 research points across development tracks. Planet components are just to support research needs? Wants components that decrease build time/costs. Wants to have factories everywhere so it can reinforce easier. Rarely excavates past 1st level.
    • Expansion – wants all planets equally
    • Diplomacy – Won’t engage in diplomacy with neighboring players. Will only ceasefire otherwise
    • Minor Factions – Doesn’t go out of its way to use them. Will conquer them once they are on its border.
    • Culture – spreads culture proactively (wants the income/resource boosts)
    • Assumed weaknesses – It prioritizes expansion at all costs. It will focus on having a quantity over quality when it comes to units. On small maps and in the early game it’ll likely be a threat, but once it reaches a certain size on larger maps it’ll probably have issues defending some of its worlds or spread too then leaving core worlds vulnerable

I would imagine the current AI is pretty close to what I’d expect of something aggressive. A straightforward capture as many planets as possible as fast as possible. Sins2 gives me the impression that non-aggressive playstyles are being removed – so the aggressive AI seems fine in this case, at least so far. Playing tall isn’t viable in the current built.
-=-=-=-=-=-=-=-=-=-=-=-=-=-

  • Defensive
    • Fleet – This composition I would imagine wants to win via battle of attrition?
    • Components – Uses defensive ones
    • Defenses – will protect all worlds via Starbase & planetary shield. Early game, while waiting for research to unlock fleet supply will build turrets & repair structures. Wants to build & add components to its Starbases often? Possibility of it using a unique formation for defensive structure placement?
    • Research – Will unlock fleet supply and new ships/structures. Will research ability to colonize planets. Will research hull/armor/shield/repair upgrades
    • Planet development – Will attempt to max out defensive planet tracks after a single point in the commerce & mining planet tracks. Prefers defensive planet components. Often excavates.
    • Expansion – wants all planets equally
    • Diplomacy – Will ceasefire as much as possible. Also will backstab neighboring players one at a time (attempting to limit number of enemies to 1-2)
    • Minor Factions – Will never conquer minor factions.
    • Assumed weaknesses – It doesn’t focus on expansion so it’ll be easier to contain. It’s planets will be harder to push and because it’ll have a smaller empire it should be able to use its fleets easier (less distance to travel)

I’m really not sure how this AI playstyle can be realized in the current build. So far I haven’t been able to build a defensive-type approach that’s viable late game. Defensive structures don’t seem to be a threat against more than 100-200 supply and that assumes no capital ships. Maybe if the structures targeted the support cruisers & non-capital ships first so far as target priority goes, but eventually they’ll have to target the capital and then likely end up ignoring the reinforcement units. The only defensive aspect I’ve seen work well is garrisons but that’s because they’re comprised of ships instead of structures.

I can’t imagine a defensive AI being viable, at least not in the current version of the game (1.21.12). End-game fleets walk through Starbases/planet defenses as if they weren’t even shooting back. Note that these are just generic AI fleets with a handful of cap ships & maybe a titan (1500 supply, sometimes more). The only thing defensive structures seem good for at this point is to abuse the AI (it thinks these structures are more dangerous than they are and will postpone attacking). The only case that might have a shot would be a defensive TEC Rebels (Novalith cannon + culture raids), though I think that goes against the TEC rebels philosophy.

If defenses were to get some kind of utility given to them, or perhaps a mid-late game research that reduces their slot cost by 1 (just so attacking fleets can spend an extra minute grinding up the turrets) then this AI type might become more viable? Aside from TEC Loyalists using garrisons and the extra starbase, I’m not really sure that a defensive AI can serve as an interesting opponent especially since Sins2 seems to be strongly biased in favor of being incredibly aggressive.

Superweapons aren’t really a thing at this point & I’ve always equated superweapons with defensive playstyles.
-=-=-=-=-=-=-=-=-=-=-=-=-=-

  • Economic
    • Fleet – For flavor, would it lean into carrier fleets or the spamming of corvettes/frigates? Maybe it wouldn’t make use of many capital ships or titans in favor of sticking to the “mass produced” units.
    • Components – I would assume a bias towards components that increase income, then a random mix of offensive & defensive.
    • Defenses – Several turrets per planet at most.
    • Research – Will unlock fleet supply and (select?) ships/structures. Will research ability to colonize planets & those upgrades. Will research economy upgrades.
    • Planet development – Will attempt to max out defensive planet tracks after a single point in the commerce & mining planet tracks. Planet components favored for influence purposes. Will excavate planets
    • Expansion – Wants all “rich” planets. Favors planets with extra income type planet bonuses
      • Vasari Economic AI would want to expand to planets with better phase resonance, leading them to want to colonize the center of the map first
    • Diplomacy – Will ceasefire as much as possible. Will attempt to trade for “rich” planets as planet swaps. Wants surplus of exotics to sell to other players. Chance of backstabbing
    • Minor Factions – Will never conquer minor factions. Will attempt to maximize use of them
    • Assumed weaknesses – Its empire will be more spread out and it probably won’t have the most ideal units in its fleet. It’s good at spamming which means that AoE and crowd-control will chew through it)

The current placeholder AI only seems interested in floating resources, at least when I check to see what it was doing via game replays. For “Economic” I can see the AI wanting to maximize its income as much as possible, but floating resources just for the sake of doing so seems… odd – I’d assume the intention behind floating is to further reduce the AI’s difficulty.

When I see “Economic” I assume that this AI playstyle is intended to focus on diplomacy (business), efficiencies and income rates. This AI will want to pick its battles and rather than control all planets it wants to control “rich” planets – so a bias towards planets that have bonuses. Strike craft are “free” units in a sense; so are minor faction fleets.

Similar to the Defensive playstyle, I don’t think an Economic playstyle is viable in the current build. At least I can’t envision one.
-=-=-=-=-=-=-=-=-=-=-=-=-=-

  • Research
    • Fleet – Prefers capital ship heavy compositions with some support ships? Will actively attempt to capture as many derelicts as possible
    • Components – Favors utility
    • Defenses – Several turrets per planet at most. Homeworld & artifact worlds would see heavy defense prioritized.
    • Research – Will unlock fleet supply and (select?) ships/structures. Will research ability to colonize planets & those upgrades. Will research economy upgrades.
    • Planet development – Will attempt to max out research planet tracks after a single point in the commerce & mining planet tracks. Planet components favored for research purposes. Will always excavate planets
    • Expansion – Wants all artifact planets. Wants planets with bonuses that give extra research points
    • Diplomacy – Will ceasefire as much as possible. Will attempt to trade for artifact planets as planet swaps. High chance of backstabbing. Will always want to be an enemy to any player controlling an artifact world.
    • Minor Factions –Will attempt to maximize use of them, but might destroy them if they’re on the border
    • Inherent weaknesses – It doesn’t focus on expansion so it’ll be easier to contain. It’ll have a smaller fleet initially due to leaning heavily into research.

The Research AI could be set up to try and optimize how it develops planets – basically give it the illusion of looking smarter. For instance, a planet bonus to ship production/cost becomes a factory world. A planet bonus improving orbital extraction would see a focus on orbital structures. Basically the pursuit of more bang for your buck.

Researching of hull/armor/shields/weapons seems like it is really only impactful on capital ship heavy builds. The rank-n-file ships don’t see to really benefit from the upgrades nearly as much. Quantity is more important than quality so far in Sins2. So for an AI that really wants to be ahead in research, they’d have to lean into ships that benefit from it (namely capital ships & titans, as well as some support cruisers)

Similar to the Defensive & Economic playstyles, I doubt the Research playstyle is viable in the current build. Especially since the research playstyle tends to want to turtle before expanding – and turtling seems to be a no-go once fleet supplies reach around 1000
-=-=-=-=-=-=-=-=-=-=-=-=-=-

In all playstyle cases, fleet supply would be important when it comes to research priority – RTS is still the foundation. Despite having different research/upgrade priorities, the end of a long game would see all AI eventually research all research topics and complete all planet upgrades – the playstyle selection would be more of a way of tuning an AI’s priorities or bias towards certain things that during the early/mid game/early-late games?
-=-=-=-=-=-=-=-=-=-=-=-=-=-

Should the playstyles be used to have the AI build different fleet compositions? I guess, what do these settings actually mean? How are they imagined to change the AI?

sins ii feedback
13,280 views 4 replies
Reply #1 from Sins2 Forums Top

2] Difficulty

Context: I haven’t played on difficulties aside from medium (which is the default). I’ll get around to it, but have mostly delayed until the game systems & balance are closer to “finished”. In sins1, I ended up having to use lower difficulties because I like playing on the large maps. The game would come to a halt very fast with 9 AI having maxed out fleets/structures everywhere. This led to me lowering the difficulty so that the games would last longer.

What are the aspirations for the impact AI difficulty has? Will AI actions/behaviors vary across difficulty or will the only changes be bonus income?

The kind of factors I wonder whether difficulty will impact:

  • Structure Placement
    • Easier difficulties could see structures placed randomly in gravity wells
    • Harder difficulties could see structures placed in formations in gravity wells
  • Components
    • Easier difficulties rarely use components
    • Harder difficulties always use components
  • Influence (auctions & minor faction abilities)
    • Easier difficulties rarely use influence
    • Harder difficulties always use influence
  • Combat/Engagement
    • Easier difficulties don’t account for defensive structures/fleets
    • Harder difficulties will attempt to bypass confrontations in favor of hitting weaker worlds. Ideally the AI would be able to recognize if garrisons are in-range: the AI could either use a feint to draw away the garrisons with a bait fleet so that main forces can take over a well before the garrisons can backtrack. But this is also likely asking a bit much of the AI
  • Minor Factions
    • Easier difficulties won’t attack minor factions
    • Harder difficulties will see AI conquer some/all minor factions
  • Fleets
    • Easier difficulties use 1-2 fleets
    • Harder difficulties use multiple fleets or control groups (such as planet bombers or torpedo cruisers being sent to planets with static defenses but no fleet
  • Culture
    • Easier difficulties neglect culture or establish it reactively
    • Harder difficulties will be proactive about establishing culture
  • Diplomacy
    • Easier difficulties less likely to backstab
    • Higher difficulties more likely to backstab

Will the difficulty setting interact at all with the AI playstyle selected? Things like Aggressive AI getting extra fleet/cheaper ships, defensive AI getting extra & cheaper defensive slots/Starbase slots, economic AI getting extra logistics slots/income, research AI getting cheaper/faster/better research?

Reply #2 from Sins2 Forums Top

3] Faction behavior “overrides”

Are there plans to enable the AI to make use of the more unique aspects of the playable factions. The examples below aren't exhaustive, but hopefully they flesh out what I'm asking.

  • Functional
    • Vasari AI to build phase gates at every planet
    • Vasari AI to build ‘attack’ starbases that will follow fleets which are attacking enemy worlds (higher difficulties)
    • Build Starbases and planetary shields at homeworld
  • Lore
    • Vasari Alliance
      • Never attacks minor faction homeworlds
      • Prefers influence components for planets/units first
    • Vasari Exodus
      • Possible to see it go fully mobile (or only keep artifact worlds, or allow for either case to occur?)
      • Builds Debris reclamation centers every 3 planets apart (or as well as it can)
      • Will use their titan’s create-phase-lane ability
    • TEC Rebels
      • Prioritizes alliance with Pirates.
      • Builds Human Doctrine centers spaced every 3-4 jumps apart (as well as it can anyway)
    • TEC Loyalist
      • Lean into putting a garrison on every planet it owns & a starbase with a construction bay component. Either through “saving” a slot on each world for one or for it to retool existing worlds to install them via some kind of periodic trigger?

I'm just looking to see if there are plans to have a defensive Vasari Alliance AI behave differently from a defensive TEC Loyalist, for example. Will the AI be able to build garrisons & set them to roam, hold or defensive roam based on the setting? Will higher difficulties be able to toggle through garrison settings (hold till garrison supply threashold, then go back to offensive/defensive roaming)? Hopefully my examples give the gist of what I'm asking

Multiplicity - Control multiple PCs using one keyboard and mouse
Reply #3 from Sins2 Forums Top

4] Possibility of scripted AI?

Will the AI be modable, even if no AI editor will be made? Can it be modded in terms of weighting build order items, adding tags, or will it be able to follow a scripted build order while making use of the default combat? How much leeway could we have when it comes to tweaking it? 

  • Minor Factions (the main use the devs could have for creating a foundation for AI scripts)
    • As for how the devs would be using scripted AI, I assume it’d most likely be done for the minor factions. They aren’t players so they don’t need to be fully fleshed out. While I still want them to be able to make use of the playable race AI, having them upgrade fleet supply or research upgrades based on triggers tripped by player activities (such as researching particular items or having particular fleet supply) or based on game time would be handy to make use of.
    • For some examples:
      • Pranast United would have a watered-down tech tree that it uses to research upgrades to its hull/armor/shields/defensive structures as the game progresses. It would build its own unique components for its units. If adjacent worlds become owned by any player/npc, it'll start building up its defenses/fleet ("anticipating" an attack)
      • Pirates & the Aluxians could research higher fleet supply so that they don’t immediately knock out a player with an unlucky spawn, but they’d be able to scale as threats the longer they’re left alive (Aluxian unique research would be to increase the number of jumps it’ll travel from its homeworld – things that don’t matter in small maps but can liven-up the super-late game on massive maps
  • PvP (hardcoding of build order, research completion, fleet composition & trading/donating of resources)
    • This is something I’d assume to be made by the PvP crowd. Basically, the AI would follow a specific build order and engage in specific timing attacks. Because of the nature of this though, this would be more of a training bot as per the PvP crowds’ needs. I imagine this would have to be done on a per-map basis. For example; Shruikan, Twin Giants, Ashred & the random 2-player map will likely all have faction-specific & map-specific AI.
    • This is not something that I’d expect the devs to go out of their way to create themselves, but rather they’d allow for AI to be modded by players who want to test out strategies against specific builds/timings. I’m not sure if the tactics would be of much help, as AI does have limitations as opposed to players. Still, this could be a tool
Reply #4 from Sins2 Forums Top

As far as research goes, I believe they are going back to using the stations for research instead of planets. This is going to make me shelve the game until either an update comes out or the game is released as I don't want to get used to playing it one way (planets used for research) and then have it revert to default (spamming stations to get research tree up).

Personally, I prefer the new way with planets as it helps free up valuable orbital slots.