AI Patch Suggestions

[Edwin P.]

This will be a list of suggestions for improving the AI gleaned from my experience and other posts on this forum.

1. AI does DOW Denmark, Benelux and France well. But it does not appears to attack other other nations other than Yugoslavia.

2. Game: vs Axis AI - HQ Problem. In one game I landed an Allied Force in Greece, conquered Greece and moved north. Axis AI operated 3 HQ units, Army Unit, Tank Unit and Parachute Unit from Russian front to stop advance. At most it should have operated in only 1 HQ Unit.

3. Game: vs Axis AI - Lone Corps takes Berlin. In this game I sent a lone motorized corps (level 2) through Enemy lines and into Germany. Corps reached and unguarded Berlin and took it. AI should watch the map of the bottom of the screen that shows the limit of Allied Advances

4. Game: vs Axis AI - Russian forces cut off about seven units NW of Riga. For many, many turns these seven had been trying to take Lenningrad.

The AI needs to know when to give up an attack and move on, leaving only a blocking force in place. AI also needs to know when to pull back units.

5. Game: vs Axis AI - AI needs a better defense of France. It left only 3 units to defend France, no reinforcements sent. No reserve available.

6. The Known Allied D-Day Issue

7. Game: Allied AI should sometimes decide to use the Bomber Defense of Egypt, as mentioned by hellraiser - ie Bomb Ports so units can't be transported to North Africa. - Bomb City so units can't be operated to North Africe. This also applies when launching D-Day. Bombing Cities in the Area will keep the Germans from Operating Units in.

8. Reclaim Tech Chits - AI needs instructions on when to reclaim tech chits. I.e. When reached maximum research level in an area (this should be automatic) or at other times. Example - When Berlin is threatened, or to reclaim Russian rocket chit at start of game as Hellraiser suggests.

9. AI will never use Diplomacy as much as a Human and if it does it will not know to take Gibraltar or move Spanish fleet into the Atlantic.

10. Allied AI will not take unoccupied Norwegian Towns from Human.

11. Allied AI will not take unoccupied Brest until 1943.

12. Allied AI will not move to secure North Africa or once North Africa is secured move to take Iraq and Iran (Iran is totally unguarded! no units guard its capital).

13. Would like to see Allied AI operate Bomber to Malta occassionaly or send French Navy to Med to support British Navy in Sinking Italian Fleet once Italy enters the War.

14. AI apparantly does not retreat ships or units from Egypt through the route around the Horn of Africa or send reinforcements to Africa via this route.

15. Naval Warfare needes to be improved - as mentioned in other posts - as AI though much better than SC1 in this area - sends ships to attack piecemeal (i.e. 2 at a time) and not in large task forces. Sometimes withdrawing 2 from battle and sending 1 into battle in the same turn when it should either attack with all 3 or withdraw all 3.

More to come.

16. Game: Axis vs Allied AI. Sent lone Romanian armor into southern Causcause mountain area. AI committed 5 units to stopping him - including HQ operated from Front, 1 Armor and 3 Corps. The Romanian Corps sat on a Mine and the AI corps kept attacking him one at a time, attacking, pulling back and waiting a turn, then attacking again suffering damage but giving no damage to the Axis unit. AI should have had all units surround and attack the Romanian Armor unit, instead of guarding the surrounding mine and two cities. (this was in the SE corner of Russia).

In general, AI appears to overreact to units that are behind its front lines. This overreaction weakens the main front.

17. AI purchase decisions are too random. AI should decide on a strategy and then select the plans; planning, production, diplomacy and research, that implement this strategy.

Example: Plan - Heavy Industry for Russia

- Research Production and Industrial

- No research for Rockets, Bombers, Air, etc.

Example: Plan - Corps Flood for Russia

- Produce Lots of Corps, Research Infantry to Level 3

Example: Axis Plan - Spanish Conversion

- When France falls purchase 5 chits in Diplomacy: Spain for Germany, and 3 Chits in Diplomacy: Spain for Italy.

- Italy Researchs Ground Laying Radar 2 chits

- When Spain joins Axis then take Gibraltar with Spanish units

- When Gibraltar is taken send Italian Fleet into Atlantic to sink Royal Navy.

- Defend North Africa Plan

Example: Axis Plan - Turkish Alliance & Russian Invasion

- When France falls purchase 5 chits in Diplomacy: Turkey for Germany and 3 Chits in Diplomacy: Turkey for Spain

- Germany Researchs Motorization

- Germany sends 2 units to Turkey once it joins to lead advance into Caucaus Mountain Region

Example: Axis Plan - Turkish Delight & Iraqi Conquest

- When France falls purchase 5 chits in Diplomacy: Turkey for Germany and 3 Chits in Diplomacy: Turkey for Spain

- Germany sends 2 units to Turkey once it joins to support conquest of Iraq.

18. AI needs counter and to do this also.

Originally posted by Chris G:

Never mind supply in Africa. As Axis you just take Egypt in 3 moves no matter any supply.

1. Make two corps amphibious in Italy

2. Move them to Turkeys south coastline

3. Land one Sinai and then send the other

thru Suez and take Cairo.

Counter: Position Ship to Guard Suez.

19. North Africa - AI needs defend and conquer strategies for North Africa for both sides - allies and Axis.

20. Axis Malta Attack Plan from Terif:

To take Malta is pretty simple:

- first you need to hinder allied supply to Malta so it can´t be reinforced any more (and also not evacuated )

- to do this: bomb the fortress out of order with your italian ships and the port with AFs and/or ships

- So the allied defender has only a supply of 1 left and can be killed by the up to 4 axis AFs from Sicily (2 turns to unentrench, then kill it.

21. Mediterranean Theature - Axis AI

Basically the Axis AI needs to decide what to do in this theature. Currently it does nothing.

Example:

Plan 1 - Defend Current Holdings

Plan 2 - Take Malta

Plan 3 - Take Vichy North Africa

Plan 4 - Take Egypt

Plan 5 - Cairo Surprise - one amphibious unit takes suez and another follows in the same turn to take Cairo.

Plan 6 - Withdraw from NA to Europe

Thus it may decide to do:

Plan 1

or Plan 2 and Plan 1

or Plan 4

or Plan 2 then Plan 4

or plan 5 followed by plan 4

[ April 17, 2006, 10:06 AM: Message edited by: Edwin P. ]

[Desert Dave]

More to come.

Appreciated.

And,

Thanks for the comprehensive compilation.

[Edwin P.]

22. Game - The Axis AI Leningrad Attack is really a game breaker if the Russians have one unit in Leningrad and another to the north of it. They just keep reinforcing each other and the Axis AI ties up 2 or 3 HQ units and five to seven Axis units trying to break then down. Either Attack them right by using Air Power to prevent them from Reinforcing or move on and leave a blocking force of corps behind, and the AI should not asisgn more than 1 HQ to support the units attacking Leningrad.

23. Game - Axis AI DOW on Russia, I have not seen it wait until Russia prepares for war yet. Perhaps it should sometimes.

24. Game - Axis AI advanced slowly in Russia not taking 2 open mines that were left undefended for many turns.

25. AI French Defense from Terif AAR - transport 2 overseas French Corps to France to aid in its defense.

26. AI French Defense - France should probably not spend money on research or diplomacy in the 1939 scenario.

[ April 17, 2006, 08:59 PM: Message edited by: Edwin P. ]

[Cheese Panzer]

8a. AI Allied Tech chits:

Played out a game well into 1945 (someone had to invade America!). At the end of the game both the British and the Americans had max levels in IW and AT AND still had 5 research chits each (!) in both techs. That's a total of 2000 MPP's that weren't doing anything for who knows how long.

[Blashy]

Excellent stuff Edwin, keep it coming, I'm sure Hubert will love this thread for the AI.

[Edwin P.]

27. Game - vs Axis AI.

Russians cut off large numbers of Axis Forces NE of Riga and in the Balkans (again). Essentially motorized Russian corps were used to cut off and isolate Axis Forces. They swept behind Warsaw and created a line from the Baltic south to the Adriatic. Slow moving Armies and Tanks moved in to finish them off.

Problem was that in Response to UK invasion of Greece and Russian Advance towards Romania AI operated many units to Balkans, including 3 HQ units.

And NW of Riga 3 HQ and numerous Corps and Armies were cut off.

Solution:

AI should enforce the 1:6 HQ Rule for Germans and should withdraw forces towards Warsaw/Berlin, not towards Bulgaria or Riga.

Somehow the AI should know when it is about to be surrounded and operate units out. The trick is how to do this so it does not operate out prematurely in response to Human fients.

28. Game vs Axis AI - German Navy did not engage and sink Finish Fleet when it was in the Baltic Sea. It sent one ship to attack, when that was damaged it pulled back instead of sending other warships to attack and sink Finish fleet.

29. Italy needs a Defense of Rome Plan.

30. Germany needs Defense of Berlin and Munich Plan. I mean, once Russians approach Warsaw forces should be withdrawn for the final fight. As it was only 1 Axis Air unit defended Munich, 1 Corps defended Berlin.

31. Reclaim Research Chits continues to be a problem - At game end German AI had 5 chits in Infantry when it was maxed out and 3 chits in anti tank when it was maxed out.

32. Italy did not invest any chits in research for 2 games straight.

33. Left UK,Canada and USA totally empty of all forces. AI did not invade even though USA was at war. At the least it should have taken London.

34. Axis AI is DOWing Russia too soon, all the time. This leaves it with too few units to guard France and attack Russia.

35. Axis AI also appears to be building too many German HQ units. It should stick to the 1:6 rule for German HQ units.

35. Axis Sub movements are too predictable. One always moves towards Portugal/Gibraltar. I have learned to just wait for it to come to me instead of searching for it.

[Edwin P.]

-----------------------------------------

AI Suggestions: High Level Strategic

High Level:

1. Reclaim Research Chits Automatic once Max Level is reached.

2. Reclaim Chits Plan for reclaiming research chits in extremis - i.e. Warsaw is threatened, Venice is threatened, if script defined max chit level is reached.

3. Master Plan variable that links related plans

4. Plans triggered by number and type of units spotted/destroyed/surviving - i.e. by the information on the reports screen + known unit arrivals from game start.

Example: If Allied fleet too strong, do not build subs. If Allied Fleet weak, build lots of subs.

5. Production Rules Rules with more parameters - ie HQ - 25% [Ratio to Air/Land i.e. 20] - Thus a 25% to build if less than 20% of Land and Air units are HQ units, otherwise Zero.

Example:

HQ= 80 [20](80% to build HQ if less than 20% of units are HQ units, otherwise zero)

Air Fleet= 50 [25] (50% to build air fleet if less than 25% of units are air units, otherwise zero)

6. Diplomacy Plan should allow for purchase of more than one chit at a time.

7. Plan Priority Variable should be considered so that AI determines which plan type (or plan) to run first, second and third.

Example: At game start USA should probably run Research Plan First, then Production Plan. At end game USA should probably run Production Plan first, Research Plan second, Diplomacy Plan Last.

[ April 17, 2006, 10:13 PM: Message edited by: Edwin P. ]

[Edwin P.]

36. Vs. Allied AI

Allied AI left Edinburgh and Manchester unguarded while it responded to Axis taking of London.

Thus Axis landed at Edinburgh, Corps then landed at Manchester.

37. Vs. Russia AI

When Causcaus is threatened AI sends highest level HQ to southern most Russian City where it gets cut off and destoryed. AI should keep highest ranking HQ where the main concentration of Russian forces are.

38. Russian AI Research Plan

Ideally Russia will select a general strategy to guide its game and not follow a random research and production strategy.

Example: Plan 1 - Production Tech Emphasis

Focus on Industrial, Production and Intel to Start, then switch to Infantry and Anti-Tank. No researching of Air Techs.

39. A post mentioned that if Russia DOW first then the Russian AI throws its units away on attacking the German line. Perhaps sometimes, in this case the Soviet units should hold back from attacking, withdraw from the border and build up a stronger army.

[ April 19, 2006, 07:34 AM: Message edited by: Edwin P. ]

[Edwin P.]

40. UK Needs Variable Plans - its too predictable

40.1 Send UK & French Navy to Med to Sink Italian Navy

40.2 Early Russian Entry - i.e. Diplomacy chits in Russia

40.3 Counter Spanish Activation

40.4 Counter Sea Lion Strategy - AI should track how many times the Human Player uses a Sea Lion and other key strategies and adjust its strategy accordingly.

Example: AI tracks 5 key human strategies for 20 games: Sealion, Egypt First, Influence Spain, Influence Turkey, Early Russian DOW.

Example:

Sealion - 8 of 10 games - Human is 80% likely to execute Sealion, then Allied AI is 80% likely to execute Sea-Lion Defense Strategy.

40.5 Influence Russia to Enter War Early

40.6 Influence USA to Enter War Early

[Edwin P.]

41. Russia AI - Sends HQ unit - Konev to Forest Outsifde of Archangel when Finniah unit approaches. Finish unit advances into Archangel.

AI should have sent corps not HQ to Archangel.

42. Russian AI - Stalingrad Defense - Axis forces approach Stalingrad after taking Moscow. When Grozny south of Satlingrad threatened operated entrenched Tank and (level 2) and corps from Stalingrad. Left 2 HQs in place to guard stalingrad. Bad Move.

43. Russian AI - did not switch to Sverdlovsk + Chelyabinsk Defense after Moscow and Stalingrad Fell. Russian Armor advanced (Motorized Level 2) and took both.

Desert Dave - As a general rule, I would suppose that the Allied Player MUST assume that the Axis WILL attempt a Wolf-Pack tactic.

UK probably should invest in ASW right from the start.

If Axis does NOT opt for attacking the convoy lanes, then no need to spend vital MPP's to upgrade UNTIL such time as they do, if ever.

Likely not until 2 French Cruisers cannot take losses instead of the RN.

**ALSO: there is more than just the ONE way to counter them treacherous U-boots.

(Naval) Bombers, with L/R Air tech.

USA will get a (Naval) Bomber off the production queue in plenty of time to deflect some intended sub raids on American Atlantic Fleet.

Wouldn't hurt for USA to also invest in ASW neither.

Why ever PRESUME that your opponent is NOT going to go all out with a potentially VERY powerful weapon, the U-boots?

IF the Lend Lease Route is being mercilessly obliterated, well, just temporarily reduce or cease sending treasure altogether... until such time as some "counter-measures" have been established.

[ April 20, 2006, 08:21 AM: Message edited by: Edwin P. ]

[Timskorn]

Hey Edwin, have you tried adding some of these events via the editor? It seems there are a good number of scripts missing from the 1939 Fall Weiss campaign that, unless HC is putting them in with the patch, we'll have to do ourselves.

For example, I never saw any script to take Norway or Sweden for the Axis AI, nor any counter-Allied script to take either of these back. There's also no scripts for either AI in North Africa, except one for Italy to send some troops to Tripoli. Hopefully these will be in the next patch, but if not they can certainly be added with the editor.

Also, we can implement the possible AI strategies you mention in your first post and just weigh them with percentages. Would certainly be cool to know the Russian AI could be choosing between 4 different unique strategies.

A lot of other things are simple changes you can make fairly quickly, such as making sure certain cities are garrisoned or focusing production techs during certain years/months.

It sounds like HC is busy working on a lot of the basic AI scripts, but I feel it'll always be up to us to tweak the game to a point that we're happy with. I'm sure HC isn't going to focus his time making 4 variable strategies for each major nation. The exciting thing is we can add scripts that mimick real player strategies...Rambo AI.

[Edwin P.]

Yes, I am working on scripts for the editor.

In fact for Norway there are really two scripts for the Allies - 1) Take the northern town that is the source of the Convoy route if it is not guarded! and 2) liberate Norway once Germany DOW Russia and is occupied elsewhere!. I will try my hand at this on the weekend. The key is to have the Allies screen for Axis ships that may be laying in wait.

As Blashy and yourself have said, its up to the users to tweak the AI if we want it to be all that it can be.

[mustang96]

Hey guys

Just wondering are you going to make these great AI changes available to us people who have no clue how to make these changes? I personally would be greatfull!

[Timskorn]

Well, as soon as I get the AI to a point where I'd feel it's worth releasing to the public I'll certainly do so. But, Hubert will likely beat us to the punch with the patch. I'm not sure when it'll be released and how much he is tweaking the AI, but I'm assuming it'll resolve a good bunch of the issues I'm trying to address in the editor.

[Edwin P.]

Spanish Counter

From the AAR's it appears that Spain is the Key for the Axis and that Terif has a good counter to this;

1. UK uses Diplomacy on Spain!

2. UK reinforces Spain!

3. Allies Launch D-Day in Spain not France!

Of course to stop Allied Transports from landing in Spain the Axis has to bomb the Norther Port!

[Edwin P.]

French AI Strategies

{

#NAME= France - Algerian Troops Recalled To Paris

#POPUP= French Foreign Legion Recalled to France

#FLAG= 1

; Checked for Once

#TYPE= 0

#COUNTRY_ID= 2

; 50% to Occur

#TRIGGER= 50

; Level = 0 as other level checks don't work

#LEVEL= 0

#PLAN_ID= 3

; One Corps

#SIZE= 1

#LENGTH= 0

; Marseilles - Must be a port for a Transport Order

#GOAL_POSITION= 71,23

#DATE= 1939/09/03

#STEAL= 0

; Set friendly positions:

; Tunis

; Marseilles

#FRIENDLY_POSITION= 77,31

#FRIENDLY_POSITION= 71,23

; Set variable conditions:

; 1st Line - France politically aligned with Allies and not surrendered

#VARIABLE_CONDITION= 2 [2] [100] [0]

; Set tactical conditions:

; 1st Line - Tunis not tactically threatened (dummy condition)

#TACTICAL_CONDITION= 77,31 [3]

; Set dummy activate position (no units at position 0,0)

#ACTIVATE_POSITION= 0,0 [0,0] [0,0] [0]

; Set dummy cancel position (single neutral unit at position 0,0). This is not possible as no

; unit can occupy tile 0,0 so event will not be cancelled due to #CONDITION_POSITION

#CANCEL_POSITION= 0,0 [0,0] [1,1] [0]

Summary:

Units

1. Tunis Corps

- Send to France, Egypt or UK

2. Syrian Corps

- Send to France, Suez Title, Cairo or Amman

3. French Air Fleet

- Disband?

4. French Navy

- Send all Navy Ships to Central Med to engage Italian Fleet

- Send all Navy Ships on Sub Hunt in Atlantic

Diplomacy

1. Iraq to start convoy routes to UK or

2. Sweden to stop convoy routes to Germany or

3. Do nothing

[ April 23, 2006, 01:32 PM: Message edited by: Edwin P. ]

[Edwin P.]

Question - Why doesn't the BEK in Brest attempt to return to England?

It stays in place and allows the Axis to eliminate it.

[Timskorn]

Cool stuff! I've used a lot of editors from RTS games like Empire Earth (even worked professionally doing scenario design), but we never had an easy way to save and transfer triggers. I just think it's awesome that we can all pool scripts together like this and simply copy them over if we want them in our version. Can't wait to see what we have in a year!

Yeah, I've noticed the BEF army never leaves port too. I haven't checked any scripts for it, but, is this something the AI should be doing without scripts? I'd also be interested to know from Hubert how the general AI works (the stuff we can't script). It would help us scripters in regards to how much or how little we should script something. For example, with the AI eventually launch an offensive from Alexandria as the UK? Or is that something that must be absolutely scripted?

[Edwin P.]

There is no script for the BEK leaving Brest, and it appears that it must be scripted.

The AI does not launch an offensive from Alexandria. I wrote a script for the Allied AI to do so if the Axis leaves Tobruk unguarded.

Currently if Tobruk is left unguarded the Allies AI bombards it with Air and naval vessals while the Italian Fleet is in the Atlantic. It never sends a land unit forward to take it.

[ April 23, 2006, 05:14 PM: Message edited by: Edwin P. ]

[Edwin P.]

From the AI NOTES Thread:

Of course I may not want to execute the send Italian Fleet to Atlantic strategy if Allied Naval Forces are too strong.

Using information from the Reports Menu a Human player can evaluate the strength (Air, Naval and Land) of his opponenet. I wonder if a condition will be added so the AI can do the same?

Example:

REPORT_CONDITION_G= Country_ID [unit type reference value][unit count from the Reports screen]- Condition where the unit count on the Intelligence Report will cancel the event. The G is for Greater Than.

Example:

REPORT_CONDITION_G= 1[2][6] : Cancel Event if UK royal naval fleet units are 6 or more. The G at the end is for Greater than.

Ideally there would be another condition for REPORT_CONDITION_L= 6[2][8] Where the A at the end is for Less Than. Cancel the plan if the Italian Navy has 8 or fewer naval units.

Thus one could have a script that would be executed - based on information from the reports screen - by the Italian AI ONLY if UK Naval Forces are Less than 6 AND Italian Naval forces are greater than 8.

This information could also be used in Research and Production Scripts.

Example: If the Axis have <2 Naval Ships and the Allies > 6, why should the Allies build more ships. If the Axis have > 12 ships and the Allies < 6 the Allies probably needs to build more.

Summary:

; For Greater Than

Intel_Report_G= Country_ID [unit type reference value][unit count from the Reports screen]

; For Less Than

Intel_Report_L= Country_ID [unit type reference value][unit count from the Reports screen]

Example:

Intel_Report_G= 1[2][4] ; US has Air Fleets of 4 or more

; LAND= 1

; AIR= 2

; SEA= 3

Of Course, there should be scripts for the AI to respond to the spotted Tech Level of Enemy units.

[ April 24, 2006, 11:51 AM: Message edited by: Edwin P. ]

[Timskorn]

Yes, I'd love to see scripting like this. I'm so used to having this kind of control in editing. It makes sense to be able to switch AI research on the fly, for example, if they find out the Germans have level 4 subs. I'd like to be able to script the ability to switch research over to compensate for this.

On another note, I think I've found why the Russian AI has been acting so newbishly. I've done a bunch of changes and plan on testing them soon, but here's the gist of it:

In the USSR Garrison scripts, they are set to 'loop' until the Cancel position condition is triggered. For all the USSR scripts, these are:

1. If Axis is 1 tile away from (city).

2. No Axis units within range of (city)

So once a single Axis unit reaches a tile outside the city, the garrison script stops. But the BIG thing is #2. The AI will never 'dig in' their lines to the rear because the Garrison script is only meant to activate when the AI is getting close to the city (#Activate condition). The scripts activate the Garrison script, typically, when 2-4 Axis units get within 3-4 tiles of that city. So they don't have time to dig in.

I've redone the scripts so front-line cities like Riga and Minsk aren't garrisoned via script. Also, cities like Voronezh, Moscow and Stalingrad (and their tactical surrounding locations) get garrisoned as soon as the 1st German units enter into Russian territory. This should give them time to 'dig in' a proper defensive line well before German units reach them. I still have to test it, and I'll post an example script of what I have for Stalingrad. The Garrison is triggered when 2 Axis units are next to Moscow, and it is cancelled once 3 Axis units are adjacent to Stalingrad:

; Garrison script

{

#NAME= USSR Garrison - Stalingrad Defensive Line

#POPUP=

#FLAG= 1

#TYPE= 2

#COUNTRY_ID= 4

#TRIGGER= 100

#LEVEL= 0

#PLAN_ID= 0

#SIZE= 1

#LENGTH= 1

#GOAL_POSITION= 127,16

#GOAL_POSITION= 126,16

#GOAL_POSITION= 127,17

#GOAL_POSITION= 125,15

#GOAL_POSITION= 126,18

#GOAL_POSITION= 124,15

#GOAL_POSITION= 125,18

#GOAL_POSITION= 123,14

#DATE= 1942/03/03

#STEAL= 0

; Set friendly positions:

; 1st Line - Stalingrad

#FRIENDLY_POSITION= 127,16

; Set variable conditions:

; 1st Line - USSR politically aligned with Allies and not surrendered

#VARIABLE_CONDITION= 4 [2] [100] [0]

; Set tactical conditions:

; 1st Line - Sverdlovsk not tactically threatened (dummy condition)

#TACTICAL_CONDITION= 135,5 [3]

; Set activate position

; 1st Line - Axis within range of Moscow (About to Fall)

#ACTIVATE_POSITION= 112,10 [1,1] [2,2] [1]

; Set cancel position

; 1st Line - Axis within 1 tile of Stalingrad

; 2nd Line - No Axis units within range of Stalingrad (disabled)

#CANCEL_POSITION= 127,16 [1,1] [3,3] [1]

}

[Timskorn]

I haven't tested this yet either, but I think there may be a way to make sure the Siberian reinforcements aren't snatched away and separated. I looked under the unit.txt file in the Events folder and located the conditions for which the Siberian transfer takes place. I think if we set up an Offensive_Plan around those exact coordinates and make it "wait" for the oppropriate # of units, it would activate the plan and group those units into a single offensive group to throw back at the Germans.

[Edwin P.]

Excellent idea!

[jon_j_rambo]

It's impossible to program an A.I. for a game with so many possibilities.

[Timskorn]

Check out Negative Nancy here.

We're just trying to add some stiffer competition, not the next Deep Blue.