Gran Turismo 6 AI discussion

  • Thread starter JBanton
  • 803 comments
  • 83,068 views
Maybe some of you have this problem in GT5 as well (In Arcade mode): First corner of Nurburgring (the circuit or TypeV/24H), some of the AI brakes so early and some don't. So I brake early to not hit the guy in front, but then the guy behind me comes full force and hits me in the back. I usually have the strong damage on, so this makes me restart the race every time.

Another thing is that the AI always takes the same lines through the corners, and they go off the track at the exact same spot, every lap. It's to the point that I just plan ahead for the particular corner and wait for the AI to do the same thing they always do, then pass. This becomes very uninteresting quickly. It would be nice if AI did something different lap after lap, or adapt to the surrounding (which they sometimes do, if I'm putting pressure on their tail for some time, they are bound to make a mistake and go off the track).
 
Well, we haven't seen the final AI yet. At least Kaz says that it was just a placeholder so far... :indiff:
 
No they patched it because it was literally on rails, you could park your car in the middle of the road and they would go straight into you and instead of backing up and go around you they would try to ramm you off the road. Also, some races were extremely frustrating because of this, both in Aspec and Bspec like the Historic one with the 2j and the Toyota 7. I'm sure most of us here won that one with either a Countach or a 512BB but we struggled and don't even get me started on the Bspec version of the same race.

Imagine how hard and frustrating this races were for newcomers and casuals, let's not forget that they're the ones making Sony's money.

It's not what I experienced. True, some races were hard, but I found the AI to be more aware than you state. Might just be my personal experience with the game pre-patch. But it was definitely a lot better than post-patch, where the AI would almost bow for you when they move aside to let you pass. That behaviour would be fine on lapped cars, but on cars you're actually racing against it completely destroys the game.
 
... That behaviour would be fine on lapped cars, but on cars you're actually racing against it completely destroys the game.

Quoted for devastating relevance and truth.


Can somebody tweet Kaz and let him know that A.I. is very important? It kinda seems like PD doesn't think A.I. is a priority.

...ok, it REALLY seems that way.
 
For me, the AI is the most frustrating thing in GT5. I can just about put up with the poor racing lines and moving roadblock scenario that sometimes happens (not always to be fair) but the fact they are so slow, uncompetitive and brake so early just kills any enjoyment.

Its also frustrating that you pick a slower car to try and force a good race which you get a lot of the time, but the game still throws in a car that is super quick that you have no chance of catching.
 
AI go off the track at the exact same spot, every lap.if

Yep, that happened even in GT1 (Clubman Cup - Autumn Ring and Grand Valley)
and in GT2 they really liked to spin off at the first (triple) corner of Apricot Hill.
I wonder if in GT6 they will be making the same mistakes at this track :yuck:
 
Besides to also throw it out there, every car enthusiast who owns a Playstation system buys Gran Turismo. Not only do we buy GT and most likely the digital content offered online because we support the franchise. But we also buy every other driving game too. There's going to be no such thing as "putting a dent" if we buy every racing game anyway.
:)
rofl.gif
 
Indeed. I hope that if GT6 is a poor release it opens the door for other developers to really put a dent in PD's sales.

That's given me an idea for an interesting social experiment, based on the above prototype:

Judging a game by some set of subjective standards chosen by some arbitrary customer and wishing some kind of negative outcome upon the game's producers should it fail to live up to those expectations chosen by the individual.

Judging a person (e.g. via their contributions online) by some set of subjective standards (by exposure to an arbitrary audience of my choosing) and seeing what outcomes are wished upon the person according to the expectations of the audience I choose.


I personally hope other games can thrive irrespective of GT's presence, assuming they're good games. There's plenty of room in the marketplace, because we don't know its boundaries yet. There must surely at least be room to fill in the gaps that GT leaves (intentionally or otherwise) in terms of what we already know is possible.
 
I'd note that SG did say if it were a poor release. ;)

I largely agree with you though, Griffith. I do enjoy that GT has competition these days, as I'd assume it encourages the team to not rest on their laurels when some killer feature from another game gets a ton of recognition. Video game sales are quite a bit different from cars; people don't pick one at the expense of another nearly as often, so like you say, there's a lot of room in the marketplace for people to help grow the sales of other games in addition to purchasing GT.

Of course, the ridiculous idea that shows up here from time to time, that people need to pledge allegiance to one particular franchise over others, is thankfully not that common.
 
Judging a game by some set of subjective standards chosen by some arbitrary customer and wishing some kind of negative outcome upon the game's producers should it fail to live up to those expectations chosen by the individual.

I didn't care much for reading your experiment but the bottom line is that I want the best racing games available - if a developer fails to adequately improve a game on fan feedback, they will feel the need for a more urgent response if the result is a significant loss of sales.

I personally hope other games can thrive irrespective of GT's presence, assuming they're good games. There's plenty of room in the marketplace, because we don't know its boundaries yet. There must surely at least be room to fill in the gaps that GT leaves (intentionally or otherwise) in terms of what we already know is possible.

I think there is room to fill in the gaps, but I think the most apparent impact of GT's market presence is a hesitation from developers to incorporate the features that overlap.

I think this is what Codemasters has attempted with Grid 2. They have Touge, drifting, (basic) livery editor, which is filling the gaps so to speak. However it seems to me like the arcade structure/physics of the game is how the developers have to distinguish themselves from the GT series to maximise their sales.
 
Last edited:
I didn't care much for reading your experiment but the bottom line is that I want the best racing games available - if a developer fails to adequately improve a game on fan feedback, they will feel the need for a more urgent response if the result is a significant loss of sales.

Yeah, that's fine. The point both you and Slip missed was that "best" is, at best, subjective. You have a "best", I have a "best", everyone has their own distinct "best"s. But obviously a loss of sales will makes them re-evaluate things.
I don't think you should arbitrarily wish that on PD just because they don't make the right kind of "best" for you, that's all.

My "social experiment" idea was just a light-hearted dig; nothing meant by it (except, you know, "best"; or, rather, "poor").
I think there is room to fill in the gaps, but I think the most apparent impact of GT's market presence is a hesitation from developers to incorporate the features that overlap.

I think this is what Codemasters has attempted with Grid 2. They have Touge, drifting, (basic) livery editor, which is filling the gaps so to speak. However it seems to me like the arcade structure/physics of the game is how the developers have to distinguish themselves from the GT series to maximise their sales.

You may be right about that, at least for giant productions like GT and GRiD etc. I think there is more space for smaller projects, personally.
 
Yeah, that's fine. The point both you and Slip missed was that "best" is, at best, subjective. You have a "best", I have a "best", everyone has their own distinct "best"s. But obviously a loss of sales will makes them re-evaluate things.
I don't think you should arbitrarily wish that on PD just because they don't make the right kind of "best" for you, that's all.

It is subjective, but my main criticisms stem from the most common concerns raised by the GTP community (AI, sound, Livery Editor) and PD being behind other developers in these aspects. There are other concerns I have with the series, but purchasing GT6 for me will be decided by those three issues. I don't think my concerns are outlandish or unreasonable in this regard - Kaz knows that the fanbase sees those aspects of GT as a significant concern.

All concerns are arbitrary, but when there is a general consensus amongst the fanbase about shortcomings of core gameplay aspects, I would expect that these be adequately addressed one way or another. It might be selfish, but ultimately it is my gaming experience that is most important to me - not Kaz or anyone else at PD.


My "social experiment" idea was just a light-hearted dig; nothing meant by it (except, you know, "best"; or, rather, "poor").

No offense taken.
 
It is subjective, but my main criticisms stem from the most common concerns raised by the GTP community (AI, sound, Livery Editor) and PD being behind other developers in these aspects. There are other concerns I have with the series, but purchasing GT6 for me will be decided by those three issues. I don't think my concerns are outlandish or unreasonable in this regard - Kaz knows that the fanbase sees those aspects of GT as a significant concern.

All concerns are arbitrary, but when there is a general consensus amongst the fanbase about shortcomings of core gameplay aspects, I would expect that these be adequately addressed one way or another. It might be selfish, but ultimately it is my gaming experience that is most important to me - not Kaz or anyone else at PD.

And all "general consensus" does is lead us down lowest-common-denominator alley. I'd best hope that market opens up a bit, I suppose. But now we're both being equally selfish. ;)

I don't mean that there aren't things I want to see fixed in the series. I just don't personally believe that every game has to pander to "general consensus" at all, because none of us actually agrees on any total image of the game if you pay close enough attention. Or rather, the only way to do it is to create an explosion of options; which is a balance, bug-finding and usability nightmare!
No offense taken.

:cheers:
 
I don't mean that there aren't things I want to see fixed in the series. I just don't personally believe that every game has to pander to "general consensus" at all, because none of us actually agrees on any total image of the game if you pay close enough attention. Or rather, the only way to do it is to create an explosion of options; which is a balance, bug-finding and usability nightmare!

Well, each genre has it's list of things that could be considered fundamentals, it's an enormous faux pas for a game not to include these things. And these fundamentals change over time, just as expectations for things like graphics evolve over time.

A company can skirt some of these fundamentals, but the results have to make up for any perceived shortfall in quality.

What constitutes a fundamental is debatable, but since it's more or less defined by what is common to games in the genre it's not too hard to compare common features.
 
Well, each genre has it's list of things that could be considered fundamentals, it's an enormous faux pas for a game not to include these things. And these fundamentals change over time, just as expectations for things like graphics evolve over time.

A company can skirt some of these fundamentals, but the results have to make up for any perceived shortfall in quality.

What constitutes a fundamental is debatable, but since it's more or less defined by what is common to games in the genre it's not too hard to compare common features.

You have to wonder where these genres even come from, though, if there are lists to adhere to... :dopey:

I agree that expectations exist, I just don't think we should apply the same expectations to everything, "fundamentals" or otherwise, because, if these are or are not truly fulfilled, all you end up with is the same old game (as any other), or the same old disappointment, either way.

But I forget what this even has to do with AI, now... It could serve as an example, I guess: frankly, I like to think that we can do much better than the "expectation" for AI in the "genre" as it stands.
 
You have to wonder where these genres even come from, though, if there are lists to adhere to... :dopey:

I agree that expectations exist, I just don't think we should apply the same expectations to everything, "fundamentals" or otherwise, because, if these are or are not truly fulfilled, all you end up with is the same old game (as any other), or the same old disappointment, either way.

But I forget what this even has to do with AI, now... It could serve as an example, I guess: frankly, I like to think that we can do much better than the "expectation" for AI in the "genre" as it stands.

You make it sound like all a game consists of is fundamentals. The fundamentals are what make the game work. They're necessary for the game to be viewed as part of the genre. But it's all the other stuff that makes a game unique and interesting.

Bread is fundamental to making a sandwich. You can't make a sandwich without bread, or it wouldn't be a sandwich*. But it's what's in the middle that makes the sandwich interesting and exciting.

I'm sorry, but for a racing game the AI is bread. It's not racing if there's no one to race against. The obvious exception being if you do an iRacing and go purely online, but GT doesn't seem to be going that way.


*If anyone brings up that heart-attack-in-three-easy-bites THING that KFC made with bacon and cheese between to bits of chicken, I will punt a puppy. :D
 
I'm sorry, but for a racing game the AI is bread.

Exactly.

The problem is Gran Turismo has a real identity crisis. I still don't even know what it's trying to "simulate".

Test driving a huge array of cars ie. the real driving simulator? No. The majority are locked in Arcade Mode and must be unlocked through, you guessed it, races!

A authentic race weekend? No. No free practice, no qualifying, no flags, chase the rabbit style "races", lackluster AI, rolling starts that aren't even rolling starts.

An "arcade" game? No, the physics are among some of the best on consoles, with more improvements coming for GT6.

In fact, other than the physics, there isn't much relevance to real world racing at all (something Kaz would be well aware of). So maybe it really is trying to be a "driving simulator", which then begs the question of why you'd force players to go through races to unlock cars to then test drive, especially when the races don't really do Gran Turismo (or real racing) justice.

For me, Gran Turismo is at it's (almost) unbeatable best when hotlapping round the full weather, 24h Nurburgring. Racing, as it stands in Gran Turismo, just has too many flaws to be worthwhile.
 
I think this is what Codemasters has attempted with Grid 2. They have Touge, drifting, (basic) livery editor, which is filling the gaps so to speak. However it seems to me like the arcade structure/physics of the game is how the developers have to distinguish themselves from the GT series to maximise their sales.

This is why I take GT graphics above anything else regarding console car games. No matter what other games have or do for the driving experience, I always come back to GT. Hopefully, eventually, soon, the AI in GT gives us something to cheer about.
 
For me, Gran Turismo is at it's (almost) unbeatable best when hotlapping round the full weather, 24h Nurburgring. Racing, as it stands in Gran Turismo, just has too many flaws to be worthwhile.

Gran Turismo: The Real Time Trial Simulator :sly:

That aside, I'd have to agree. Although, I think that there are some ways to make racing more of a challenge (besides improving AI) in GT6. One way they could do it is by using the time/weather tracks more often in offline/seasonal events as suggested here:

I sincerely hope they also merge time/weather-tracks with the static version, as this just overloads the menus. You can still set a time/weather-change track to constant track conditions - no need to make two out of it.

...that way it presents more of a challenge when combating the effects of nature.
Another way could be imposing even harsher restrictions than the current restrictions in seasonal events. E.g. More stock car races, using comfort tyres more often (and on more powerful cars), etc.
Honestly, if PD can use the effects of realism to a more profound effect, racing can be much more fun and challenging, unlike what it was in GT5.
 
Last edited:
You make it sound like all a game consists of is fundamentals. The fundamentals are what make the game work. They're necessary for the game to be viewed as part of the genre. But it's all the other stuff that makes a game unique and interesting.

Bread is fundamental to making a sandwich. You can't make a sandwich without bread, or it wouldn't be a sandwich*. But it's what's in the middle that makes the sandwich interesting and exciting.

I'm sorry, but for a racing game the AI is bread. It's not racing if there's no one to race against. The obvious exception being if you do an iRacing and go purely online, but GT doesn't seem to be going that way.


*If anyone brings up that heart-attack-in-three-easy-bites THING that KFC made with bacon and cheese between to bits of chicken, I will punt a puppy. :D

I'm not sure if we're talking at crossed purposes here, but it's obvious that AI is "bread" and I never said otherwise. I also get that finishing flourishes etc. can create two different games born from the same "fundamentals". The problem is that if you make everything so prescriptive, as the mainstream tends to be, then you lose most of the variety in the fundamentals and a lot of it in the flourishes. To be clear, there is plenty of room for improvement in AI in games as a whole, never mind within certain genre boundaries, real or imagined - what use is it to copy something else (meet expectations based on other games in the "genre") when what's really needed is innovation?

It's these "expectations" that are the main driving force both for change and for stagnation, weirdly - I mean, we can all imagine our perfect game, and this shapes our expectations, too, but we are still only experiential, so our imaginings are dictated by past games, and everything else.
This idea that a game belonging to a certain "genre" (i.e. ticking off the boxes) has to have certain things executed in a certain way is damaging, creatively speaking. It's only really useful in terms of classification, but that's only turned out to be a mess when new things just keep appearing every now and then: what genre is Minecraft, aside from Minecraft-like? What about Monaco, Ico, Dark Souls, Deus Ex, Elite etc.
This is why a proper review is really important, too, so you know what a game is about without having to rely too heavily on prejudices based on genre tags. They're nice short-cuts, but they're too prescriptive.

Basically, some sandwiches are more popular than others, and these are the ones that tend to get made more often, regardless of whether better sandwiches could ever exist - because people can only ask for what they know. I'm inclined to change my mind from time to time, so I'll happily look out for a new favourite sandwich should one come along.


Looking at Machschnel's post, it's clear that GT is "genre-bending" in the same way that my examples above are. It's arguably true that GT is at its best as a "time trial" (although I often remove that aspect, too: for me, it's more of a test drive for different concepts of what cars and driving can be, by trying different cars, or modifying others etc. - oh look, Vision Gran Turismo). What's not clear is whether the "bad" AI is caused by that focus or vice versa. Racing is in our nature, so we're always going to want to do it now and then, of course.

GT suffers, though, for trying to be all things to all people despite its strengths lying elsewhere, but that's simply what's expected of it.
I blame console gamers' diverse and conflicting expectations as a group, given they coveted the diverse racing games available on PC, when racing games on consoles largely only fitted the "arcade" category (the only sandwiches available).
Then GT released and subsequently offered that tantalising look into "simulation" for consolers, changing expectations, fueling the elitist PC vs. console "wars", driving the faux-realism (it has to seem realistic, but not be as difficult as that, generally) crusades in PC racing games, killing the perception of a good, fun arcade racer etc. - as well as being revolutionary from a standpoint of structure and scope.


In short:
Or it may not be as simple as that, but clearly GT was a sandwich no one had ever seen before: it came from not following the list, from practically defying expectations. I strongly believe it should keep doing that; there are plenty of games doing the cynical thing and giving us "what we want" - it's nice to have both approaches.
I still want improved AI, just not in any way I've seen before (or even imagined, to some extent), so I'd prefer it if PD didn't use other games as a check list, and did some proper exploration instead.

I suppose, to attempt to keep this constructive for those who don't like walls of text: shouldn't the AI approach our real world experience of racing, i.e. in real vehicles on real tracks, or even just online in games, rather than what other games have done with AI? Some games come close, in parts, in terms of the impression they give, but, as I've said before, that's not enough for me: it must be genuine. :dopey:
 
Taking the sandwich analogy one step further because I can't resist:sly: I don't want GT6 to be a sandwich I want it to be the Sandwich Shop:idea:. Options, options, options is what can make GT great and appeal to an even broader base. Whether you like headcheese:odd: or corned beef, swiss or cheddar cheese, pickles or hot peppers, Franks Red Hot or mustard or anything else, PD should be able to make a huge range of sandwiches, from the simple and easy to the massive and complex.
 
Options are great, but can sacrifice the focus and balance of a game. If tested well enough and carefully assembled from parts that interact nicely, you can get a great starting point to refine into something very playable all around.

It's obvious that it's going to be more work that way, but then this is a very expensive game already, so it might be worth diverting some of that budget into making the game one big collection of tweakable "systems" instead of a singular experience. :)
 
Not really been following GT6 due to F5, Whats the deal with AI are PD working on it to be great? Or are we getting the 2 laps behind GT5 AI?
 
Not really been following GT6 due to F5, Whats the deal with AI are PD working on it to be great? Or are we getting the 2 laps behind GT5 AI?

At the moment we simply have words from Kaz that it'll be better, we've seen no evidence of it in the demo builds to date.
 
And all "general consensus" does is lead us down lowest-common-denominator alley. ...

I don't mean that there aren't things I want to see fixed in the series. I just don't personally believe that every game has to pander to "general consensus" at all, because none of us actually agrees on any total image of the game if you pay close enough attention. ...

I can see your point about general consensus, however I just think that for the fundamentals there is sufficient criticism for PD to seriously re-evaluate what they are doing. That and I think that PD seem to lack common sense in some aspects of their game, but that is another issue completely.
 
You said it,but we're getting some sweet new tracks even if it is still the same artificial idiots from the GTA/Demo.
 
Back