Artificial Intelligence (AI) should be improved before release of Normandy - Agree or Disagree

[Sky_Wolf]

Artificial Intelligence (AI) should be improved before release of Bodenplatte - Agree or Disagree

[Chief_Mouser]

Do you mean done as a priority that might end up putting back the Bodenplatte release, or just a general wish that it might be so?

[Sky_Wolf]

You can read the poll as:

"Artificial Intelligence (AI) should be improved before release of Bodenplatte (even if this means delaying the Bodenplatte release) - Agree or Disagree"

[sevenless]

Sure. Everything that makes the experience more immersive should (and most likely can) be done. I´m glad which progress has already been made in the SP department with enemy planes bouncing my home airfield etc..

[Sky_Wolf]

On 06/04/2018 at 3:05 PM, sevenless said:

Sure. Everything that makes the experience more immersive should (and most likely can) be done. I´m glad which progress has already been made in the SP department with enemy planes bouncing my home airfield etc..

 

As far as I'm concerned, AI is the only thing I would say should be improved before the release of another theatre (Bodenplatte).

 

I would never argue that graphics, sound, visual effects, flight physics, damage modelling, etc. should be improved before subsequent releases because they are great. Most people agree with this. But the AI is not so great, and so far the poll shows that most people agree with this. The AI is usually good, but it is often not so good, ruining immersion. 

 

Therefore the AI should be improved before more content is released, especially now that the career mode is here and most of us are spending even more time flying with and against the AI. 

[Thad]

Salutations,

 

Indeed... the AI needs work. I was running one of my single player missions yesterday. My wingman shot up and was chasing a Lagg3 down. The Lagg crashed into the ground and blew up, unfortunately, the trailing AI wingman did NOT pull up and flew into the ground needlessly.

 

Something needs work for sure.

[Dakpilot]

I cannot really agree or disagree

 

I think the reality is that the release of Bodenplatte (revenue) is actually needed for a major AI update (dedicated AI programmer)

 

dev's have acknowledged this on multiple occasions, however any updates that can be squeezed out will be gladly received,

 

although far from perfect currently, there have been many small improvements over the history of BoX

 

sorry for boring reality check  

 

Cheers, Dakpilot

[Sky_Wolf]

I'm not so sure that your belief that more revenue is needed to update the AI reflects "reality".

 

The devs have the resources to create a new BoX theatre and brand new series of games (WWI and tanks), yet don't have the resources to improve the AI? Doesn't make sense to me. 

[sniperton]

IMO the first step should be a more detailed command system. When leading two flights it's a bit unrealistic not being able to send one flight against the bombers and the other against the escorts. Or not being able to order them to attack AAA first when on a ground attack mission. AI squad mates behave as if radio had not been invented. I try to cooperate with them, but they refuse any cooperation with me, or they do it only on their own will, as if they were autists.

[Gambit21]

3 hours ago, Pericles said:

I'm not so sure that your belief that more revenue is needed to update the AI reflects "reality".

 

The devs have the resources to create a new BoX theatre and brand new series of games (WWI and tanks), yet don't have the resources to improve the AI? Doesn't make sense to me. 

 

It's not that hard.

Dedicating a guy to overhaul the AI doesn't not make inroads toward the new product/revenue/continued existence.

It's just about priorities and resources...they lost their dedicated AI guy a while back...Jason is working on correcting the programmer shortage.

[SCG_Space_Ghost]

15 hours ago, Gambit21 said:

-snip-

 

Dedicating a guy to overhaul the AI doesn't not make inroads toward the new product/revenue/continued existence.

 

 

-snip-

 

I could really go both ways on this one.

[Gambit21]

I know what you mean, it definitely helps the new product...I meant getting it out the door.

I should have been more clear.

[1./JG42Nephris]

I think it is a bit to easy just seeing the AI instead the whole game here.

The studio needs to make profit in a certain degree, otherwise the next step is last - typical buisness/project management (Prince2 ;-)

If an AI dev shall work dedicated to AI fixing (what ever that will mean) you cant say how complex it will become, as it is not just changing variables to true or false.

 

So imho, the temporary AI is no game breaker, sometimes it is maybe frustrating, but the game works and the campaigns works with it aswell.

So no need for a "hotfix" or escalation and so no need to invest man days without the sight on the whole thing.

 

Fixing the AI - yes please, but  let them stay in the their plan ( am pretty sure they will, doesnt matter whatever we keep disusswing here).

I ´d be fine if a overhaul or whatever we ll call it ,would come with BoBp, as it serves the project more than a modular quick AI fix I guess.

 

[Thad]

Salutations,

 

Just pondering the games AI situation. An opinion. Firstly, I think the most of the players in IL2 participate in single player and not in multiplayer.

 

If this is correct.... the game AI is very, very important because it controls ones wingmen performance and all of the opponents units performance parameters. As a result, the games AI truly dictates the quality of the simulated experience for the player.

 

The game AI is probably the most important aspect of the game unless all of the craft are being controlled by a human being in a online mission or campaign.

[Gambit21]

The way I see it, aside from some ground attack AI which is hampering me and needs to change as soon as possible, fighter v fighter AI faces the obstacle in the PTO, not Bodenplatte as much. This is because of the disparity in tactics between the Wildcat and Zeke.

We can't have AI Wildcats turning with the Zeroes or it will be a one-sided slaughter every time. 

 

So that's where the real challenge is - although you can make a similar arguement with regard to the 262. That AI must react differently as well. It's just that as a singular aircraft the success of the entire endeavor doesn't rest upon how it's AI fights right out of the box.

[Sky_Wolf]

Thad wrote:

"If this is correct.... the game AI is very, very important because it controls ones wingmen performance and all of the opponents units performance parameters. As a result, the games AI truly dictates the quality of the simulated experience for the player."

 

I agree completely.

 

Although improving the AI might be complicated and cost money, identifying what is wrong with the AI (how it behaves versus how it should behave) is not complicated at all: 

-When flying as leader with AI wingmen, I should be able to communicate with them and they should respond to my orders. They usually don't. 

-There should be landing orders and sequences as in IL2 1946, but there is not. 

-AI planes should not crash into the ground in unrealistic ways (i.e. when nothing is going wrong).  

 

Don't get me wrong, the AI is great at times, just not all the time. 

[=362nd_FS=Hiromachi]

14 hours ago, Gambit21 said:

This is because of the disparity in tactics between the Wildcat and Zeke.

This should really depend on the skill level of Ai. More skilled Japanese Ai pilots should stick to three man formation and utilize hit and run tactics, less skilled should break formation and turnfight. When it comes to U.S., it's not clear what should be the tactics and Ai reactions, considering they didnt really stick to hit and run in 1942. I mean, Joe Bauer told his pilots: "When you see Zero's, dogfight 'em!"

But Ace skill Ai should definitely try to use Thach Weave.

In my opinion a considerable effort would be needed to make Ai react properly to different opponents and use different maneuvers. The above is my wish, but it most likely wont happen.

 

Nevertheless, Bodenplatte also needs Ai improvements. Even Kuban does, since Ai behaves really dumb at all too many times. Silly pursuit to my airfield, dropping bombs with target ahead (bridge in that case) because lone Ju-87 flies nearby or leaving escorted bombers just mess with immersion for me.

[Ehret]

Just now, =362nd_FS=Hiromachi said:

When it comes to U.S., it's not clear what should be the tactics and Ai reactions, considering they didnt really stick to hit and run in 1942

 

The yoyos and fast downward spirals - using anything else in a dogfight against the Zero is suicidal.

[=362nd_FS=Hiromachi]

That's not what U.S. pilots had to say. Cactus pilots recommended relying on armor and so-called pincushion tactics. When under attack F4F was supposed to scissor and turn into the incoming Japanese for a head-on pass. If Zero got to your tail, pilot was supposed to cut throttle, slip, skid, and otherwise kick Wildcat around until the Zero overruns then 'pour the coal on' and 'let 'em have it.' Under no circumstances was F4F supposed to turn tail and run as it was deemed a recipe for disaster.

But of course reality is not a game, and game is not reality.

[Royal_Flight]

I'm saying no - no I don't want BoBo delayed because the devs are sorting out the AI. 

But yes, the AI does need fixing. 

 

The AI is currently borderline alright, but there are a whole host of other things that need sorted, including promised features that never made it into BoK that need to be done first.

And given that the whole point of BoBo is a bit of a quick cash grab to fund future development, then its release should be prioritised.

But definitely, the AI needs sorting out. 

[Rjel]

The AI is better but still a work in progress. I'm sure it will continue being improved.

[Donik]

I'm going to revive an old thread here simply for the fact that I want to note I would be willing to pay a reasonable sum of money to have an "AI DLC", so to speak - if that's what it takes to get some decent AI in this sim.

 

I love to fly online, but I can't pause that.  And that limits my play time greatly.  I have 100x more opportunity to fly offline but the AI is so terribly bad that often, I would rather just fly circuits and practice my landings and take offs than fight the AI in it's current state.

[=gRiJ=Roman-]

I wonder if the AI could do the Basic Flying Manouvers and even more complex ones  That would open the possibilities and be more unpredictable if this is applied to different skill level in the pilots. I mean, a greenhorn pilot can do just basic ones and an old dog could do scissors, yo-yos, etc.

Edited October 28, 2018 by PA_Spartan-

[Donik]

Yeah it should obviously scale between rookie - ace.. However at Ace (or even veteran) I would expect them to fly the aircraft to their actual strengths.  I think running in to Ace AI should be rare and a real danger to your virtual life.  Rookie, average and veteran should be the usual encountered AI with vets being able to fly, fight and flee with a reasonable level of realism.

 

I've flown too many times against Ace AI now where their top maneuver is to fly inverted in a straight line for extended period of time then split-s (i guess?) and go the other direction.  It's just really bad.  I'm no pro pilot or anything.. I've got less than 20 hours multiplayer time against humans. But from playing, and reading, and just doing the normal type of activities that someone who's interested in air combat does - I understand (in theory, usually) how to fly the planes to their strengths.  Maintaining energy, etc. With that said, I can typically kill Ace AI using an A-20 without too much trouble on a regular basis.

 

The AI just doesn't get it done and it's agonizing.  I actually feel that RoF AI has better behavior in some circumstances.  There's also WoFF AI which at least has variety and some kind of random intelligence that keeps you immersed and feeling realistic.  Faking death and running for home, etc.  JG1_Barton from YouTubehas a good video showing IL-2 AI in action.

[Jade_Monkey]

AI should be improved? YES. Before releasing BOBP? Not necessarily.

Lets not hold back the content just because AI improvements are not ready.

[MasserME262]

Although being myself mostly an SP player, I will still say "disagree". If they didnt manage to make a really proper AI in all these years, they wont be able to do it in 6 or 12 or even more months, which would delay the BoBP release maybe a little too much.

Now, yes, I would love to see the AI improved, but I'm quite happy with the current one.

[Eisenfaustus]

For me the AI is the main reason why I'm still playing other sims most of the time - because everything else this sim does so great!

 

Therefore I'd rather wait a while longer for Bodenplatte than to have all the cool stuff of BP without an immersive AI to enjoy it to it's full potential...

 

Since they already started improving AI in the last patches I'm getting my hopes up

 

However I believe there unfortunatly still is very much to be done to make the AI truely immersive:

 

1st Situational (un)awareness: You cannot sneak up on AI - they know you are there. Always. Most kills in WWII airial combat were achieved in suprise bounces - not in this game...

2nd Maneuvering: They already started improving this (hooray!) but most of the time the AI still dives for the deck and then does an infinite high speed turn in the same direction... From what I read the most common defensive tactic (Until they developed something special for a certain adversary) in most air forces of WWII was a sharp break turn to have the enemy overshoot and start scissoring to press for gun solution... The higher the skill level of the ai the more sophisticated maneuvers should be of course... Also of course this includes a wish for more effective ground attacks.

3rd Decision making: When to attack when to flee? How to flee (climbing, diving, cloud cover) How to attack (maneuvering into a favourable position before you engage); all based on mission, status, type and numbers of own and hostile aircraft, weather, ammo and of course AI skill level (with rookies making the wrong call more often!) - Current AI fights you to the death no matter what (or sometimes ignores you)

4th Pilot Error: Current AI always has perfect control of the aircraft - depending on AI skill there should be pilot errors - from not managing the engine perfectly, maneuvering at perfect angles to stalling in the heat of battle (panic after near misses) or losing control when control surfaces are damaged.

5th Tactics: While of course No. 2 and 3 already include some tactical improvements here I'm talking about formations and SOPs (especially interesting for Bodenplatte as RAF, USAAF and Lw used the finger four [which is something diffrent from the games 4 ship Vic] as standard, fought in pairs and used crossover turns - none of which is possible with current AI) This of course would also neccessitate better communications - including the possibility to either issue diffrent orders to the A/C of your flight (like one pair attack, one pair flying cover) or even recieving tasks from an AI flight leader!

 

I put these in the right order of importance for me personally.

 

If all of this was realized the immersion of this game would more than double in my book...

 

One must be allowed to dream ^^

Edited February 15, 2019 by Eisenfaustus

[Rolling_Thunder]

Disagree. Folk have paid for BoBP. AI needs attention of course but folk have spent their money. This is not DCS.

 

Edited February 26, 2019 by Rolling_Thunder

[SCG_motoadve]

Agree

[Aap]

Is there a big difference? They release BoBp content as it becomes ready anyway and BoBp is available through early access anyway, so the only difference would be if it is labeled "early access" or "released". Kind of like DCS used to be, where "beta" was a lot more populated than "release" version, because it just had more content and features.

[Heckpupper]

Do I care about AI for singleplayer use? Not one bit.

Do I care about one's ability to implement AI into multiplayer missions in any significant numbers? Hell yeah I do.

[jeanba]

I have been flying careers and I agree that friendly AI is terrible

This is nearly killing the enjoyment, please do something !

[Sky_Wolf]

Time for a new poll about improving AI before release of the Battle of Normandy...

[Sky_Wolf]

Agree or disagree

[AndyJWest]

Disagree with the premise of the poll.

 

Different people in the development team are employed for their skills in different things. The person working on 3D models isn't going to be able to just switch to working on AI instead, and so on. And the only reason the developers can afford to employ someone to upgrade the AI is because they have an ongoing income stream from new content.

[Enceladus828]

Do you mean before EA begins, or at the Final release?

[No.332.Animal_NO]

Devs have shown that they are improving AI throughout the development of the game, even the last update they did so. So what exactly do you want? What's your beef? AI in this game flies within parameters of their respective aircraft, thats more than can be said for any of the other combat flight games currently available...

Edited April 17, 2020 by No.331.Animal_Mother

[BraveSirRobin]

The AI was improved with the last update.  Problem solved!

[Eisenfaustus]

I don't understand the question - they are constantly working on the ai.

 

So what is your thesis? What should they do they aren't doing already and how would that conflict with working on normandy?

[J2_Bidu]

1 hour ago, BraveSirRobin said:

The AI was improved with the last update.  Problem solved!

In my opinion, with the last update the AI went braindead. This is probably a side effect from something else.

 

Example: an SE5 is on my very high 7. He dives all the way to my height, flipping upside down while at it, passes me flattening and heads perfect North unchanging while I fry him from his 6. Happened twice in a row on Flugpark.

 

I don't remember ever seeing this before.