Can AI be competitive at the endgame state?

[Gandalf Parker]

Originally posted by Joonie73:
I noticed that AI neglects to do some things that are vital at the endgame stage, such as:
1. build high-quality troops;
2. cast high-level summons;
3. forge high-level items;
4. protect high level commanders, etc.
They do OK in the demo Version but I am wondering if these problems may make the AI a bit too easy to beat in the full Version. Any informed views?

The AI is written to give you a feel for what it might be like to play human players. It does fairly well under certain game conditions but doesnt take alot of game parameters into account. Such as, if the AI is set on high (making it think it should be preparing for a large long game) and the independents are set on low (2 or 3 setting) then its way to easy to rush the AI.

We have already discovered and discussed ALOT (see the various AI threads) that the AI doesnt present a good showing in the demo where the 40 turn limit tends to make people set it for a fast game so they can see each other quickly.

Once the full game reachs us there will probably be maps created where certain nations are pre-set in a way that reflects the early game actions of real players giving the AI a better footing for the actions it would take after initial setup.

[licker]

Originally posted by Joonie73:
I noticed that AI neglects to do some things that are vital at the endgame stage, such as:

1. build high-quality troops;

2. cast high-level summons;

3. forge high-level items;

4. protect high level commanders, etc.

They do OK in the demo Version but I am wondering if these problems may make the AI a bit too easy to beat in the full Version. Any informed views?

I've actually seen the AI do all of those things above. Maybe they don't do it all the time, or enough of the time, but I've seen Ulm use knights and HI (along with chaff), I've seen Marg and Ulm use seasonal spirits and *buckets* of cave drakes (14 of them in one army... that was nasty...).

I've seen some magic items, though high level (depending on what you mean by high level) isn't really doable in 40 turns. And I've seen protected commanders.

What the AI doesn't seem to do well is search (or find) magic sites, their gem income is almost always terrible after 30+ turns. Perhaps they are given more gems on higher difficulty levels and as such don't feel the need to search? Perhaps they arn't scripted to do heavy searching until after turn 40? I don't know, there are many issues that remain to be resolved once we can play more than 40 turns.

Also as Gandalf pointed out, you can handicap the AI by the map settings, or you can make it more challenging. I don't think though that changing the difficulty setting of the AI has any relevance to how it approaches a map, other than by having more resources to boost its economy. That by itself will have an effect, but I get the feeling Gandalf is implying that there is more to it than that, please correct me if I misinterpret your statements Gandalf.

[PDF]

Licker,
That Ulm AI build HI is no performance : Ulm ONLY has Heavy Inf, not one light troop !
It's cool if AI summon Drakes and S.Spirits ! Does it build Earth Boots and Dw Hammer too ?

[PDF]

(deleted - dble post)

[ November 14, 2003, 16:47: Message edited by: PDF ]

[licker]

Originally posted by PDF:
Licker,
That Ulm AI build HI is no performance : Ulm ONLY has Heavy Inf, not one light troop !
It's cool if AI summon Drakes and S.Spirits ! Does it build Earth Boots and Dw Hammer too ?

The point is that Ulm had a nice group of knights and HI in his army, not just militia and LI recruited from his indie provinces. Others had said that they had seen Ulm with ~10 HI and 90 militia and assorted junk. I've seen Ulm more often then not with about 40/60 between the 'good' units and the dreck. Ulm also had at least 14 Cave Drakes running around on turn 30, eventually he pulled them all into one big army (with his HI and whatnot) to try and stop my Behemoths which were pulverizing his Infantry. I also had ~40 LI and a mix of about 40 Serpents, Swamp Gaurds, and Falchioneers. Two behemoths and a handful of other undead. I was able to get him to rout, but I took heavy casualties, and wasn't able to kill more than 2 of the Cave Drakes...

All I'm saying is that I've seen the AI use its forces wisely, there's still room for improvement (there always is) but its not as bad as some people have made it out to be. I also notice that the AI is generally quite good at researching, which may explain in part why we don't see very good gem incomes for him. Still the AI could build more forts and probably labs.

[Particle]

Originally posted by Joonie73:
I noticed that AI neglects to do some things that are vital at the endgame stage, such as:

1. build high-quality troops;

2. cast high-level summons;

3. forge high-level items;

4. protect high level commanders, etc.

They do OK in the demo Version but I am wondering if these problems may make the AI a bit too easy to beat in the full Version. Any informed views?

You should use the AI thread for this..

[Joonie73]

I noticed that AI neglects to do some things that are vital at the endgame stage, such as:

1. build high-quality troops;

2. cast high-level summons;

3. forge high-level items;

4. protect high level commanders, etc.

They do OK in the demo Version but I am wondering if these problems may make the AI a bit too easy to beat in the full Version. Any informed views?