The AI has to be the priority for Halo 5

Howdie masses! This might start on a bit of a tangent but I’ll get to my point soon enough.

I’ve been playing a lot of Forza 5 lately and its started me thinking about just what this console generation is going to be all about. Now if you’re toodling around on some other system than the Xbox One you might easily think that it’s just going to be about more online integration, fewer walls between console functions, and better graphics, but being the cultivated sort that’s had the taste and civility to buy into Microsoft’s latest cash generator I can snootily point out that’s all BS.

What it is all about (to me) is the AI. In Forza 5, for anyone who isn’t fanboy enough to have followed along with its PR or simply have bought it, the game tracks your racing data to develop an AI which mimics your playstyle. It then inserts this AI into other people’s games, and likewise populates your races with other people’s AI’s. Now the PR line is that you have “Real” people in your races, but I think this falls short of describing just what’s best about this system.

The AI is incredibly VARIABLE. You approach someone and you don’t quite know which direction they’ll move to block you, how they’ll take that turn, or in some cases even if they’re going to make it. At a stroke, IMO, it’s resulted in one of the very best games I’ve ever played, never mind that my usual stomping rounds are sci-fi shooty games or the other criticisms you can level at Forza 5 (ex. not enough Jezza). They’re all moot when such a FUNDAMENTAL shift in gameplay has taken place.

Now (this is where the above tangent is leading up to) just think of what a shooter, namely Halo 5, could do with this sort of not “real” player mimicry but this almost randomly variable AI characteristics. One pack of grunts isn’t going to respond in quite the same way as any other pack even if all other factors are held the same. Charge forward and you might have a reciprocal push back, a retreat, or an awe inspiring mix of everything running at not only the probability of each response in the AI repertoire but also how those probabilities are varied according to my proposed random-characteristic generator. Vaguely similar attempts were made in Halo 3’s development to push more variable AI behavior (by simply working at the base number of responses) but because the AI was still consistent across all encounters the variability Bungie put in there could easily be washed out by 1.) a consistent playstyle on your part and 2.) how the AI’s quite simply all died too quickly for their AI behavior to flush itself out.

For Halo 5 I hereby propose that 343 spends MOST of their effort (for the campaign but I’d say multiplayer can suck it too to bring us this upgrade) into developing a smarter (ie. more capable), AI whose “personality” is varied randomly on an individual basis, and that in some way or another can maintain presence on the battlefield through greater numbers and/or through greater resiliency. At a stroke this builds in a huge amount of not only replayability but depth because of the building levels of complexity in more dynamic AI responses across all encounters and more importantly between playthroughs.

They could keep all else equal, level deisgn, the sandbox, ect. (though needless to say these should be improved as well) and you’d still have a radically improved campaign/co-op that could easily support whatever else 343 did to Halo 5 because even something as unambitious as knocking out a copy of Halo 4’s SOPS mode could be made intensely playable with this sort of improved AI behavior. Added with somewhat random AI spawns and somewhat randomized AI equipment sets and you could easily have a game to finally put Combat Evolved to shame.

I think they need to stop making the AI more advanced and make them fun again. I still play Halo 1, not because the AI are advanced but because they are fun to fight and to fight with. The other Halo games just don’t have that fun factor with the AI in my opinion.

I agree, and with X1, they should be.

The priority should be making a good game.

Better AI is merely one facet of making the above happen. But focusing or making that facet the priority is counter productive as all the other things that make the game good will fall by the wayside.

The AI being pretty derpy in H4 is only one problem with the game. There are numerous others. Fixing just the AI won’t make the game substantially better.

By all means, update the AI. But update everything else as well.

> The priority should be making a good game.
>
> Better AI is merely one facet of making the above happen. But focusing or making that facet the priority is counter productive as all the other things that make the game good will fall by the wayside.
>
> The AI being pretty derpy in H4 is only one problem with the game. There are numerous others. Fixing just the AI won’t make the game substantially better.
>
> By all means, update the AI. But update everything else as well.

Heh, I agree with the OP. Maybe not make the AI THE priority, but definitely one of the most important ones. The AI is what can really help push the game forward. Just look at the scarab battles in H3, they were unique and really fun encounters that relied on a much improved AI. I personally think Bungie were a bit timid in Halo 3, but it’s the AI that can make these things possible. If you want huge–Yoink- battlefields, you’ll need the AI to act accordingly. If you want crazy new types of encounters, you need the AI. Basically, if you want to do almost anything that wasn’t possible before, you need to improve on the AI. The engine has a lot of say in what is possible or not, but I’m not sure it has as much impact on the ability to pull off some new ideas.

Well, pimping up the engine, no, doing an entirely new badass engine is definitely a must too. Too long has the Halo games been running on this dinosaur. It was a good engine back then, but it didn’t stood the test of time. Piling up on top of an old engine with certain key flaws will keep the same limitations. And I bet Reach or Halo 4 suffered a lot from that.

> The priority should be making a good game.
>
> Better AI is merely one facet of making the above happen. But focusing or making that facet the priority is counter productive as all the other things that make the game good will fall by the wayside.
>
> The AI being pretty derpy in H4 is only one problem with the game. There are numerous others. Fixing just the AI won’t make the game substantially better.
>
> By all means, update the AI. But update everything else as well.

But should updating the polygon count or increasing number of buttons in forge have the same priority as the AI? My argument is that AI should take precedence over other features, not that it should be the ONLY one developed though I did use that as an extreme example to illustrate the point (it’s difficult to make the case that any other single update would be able to justify a new game. They just don’t have the same potential, hence why the most effort should be allocated to exploiting the huge opportunity that lies in more fundamentally variable AI.)

> The priority should be making a good game.

And how do you make that happen? By adding +1 to all facets? You need to set priorities based on not only where Halo’s strengths are (so it can continue on having a unique competitive identity) but also where it can be most improved. If you fail to recognize the latter then the potential for Halo 5 is severely limited because the effort you spend in making the WHOLE game a superior one will be wasted on so many short tangents, while the deeper features (ex. the AI) are left relatively unexplored and thus underutilized. You may have made a generally good game, but not the [realistically] best possible game.

Now by all means 343 should do something about the graphics, do something about balance. It’s a big studio with lots of different departments working on a collective project. But among them, the most consideration, at least, should be paid to the AI because unlike such things as multiplayer balance it has the definite and achievable potential to advance this franchise beyond anything else we’ve seen.

With every next gen technology (this time, it’s “cloud processing”), everyone thinks that because of the extra processing power, AI will be smarter and better than ever. This is never the case though because the limitation on AI isn’t processing power, it’s the number of programmed responses. The reason Forza 5’s AIs seem so human-like is that it’s pulling from a larger pool of programmed responses than any other racing game (perhaps any other game period) due to the player data collection. How would you integrate such a system into Halo? By capturing data from players playing Campaign or co-op modes and feeding that into Elite, Knight, or Brute AI? And would you do the same for units unlike players such as Grunts and Jackals as well?

Another thing to consider is that racing games are one-dimensional: there is only a forwards and a backwards. Shooters take place on two or three dimensions, which means the AI must be a lot more complex and robust. The types of data and responses needed for a racer’s AI is simple compared to the types of data and responses needed for a shooter’s AI.

I agree that AI is important, but I don’t think that current AI capabilities are as good as you think they are (while keeping all other aspects of the game on par, of course).

I agree that AI in H4 was noticeably bad. I still like the campaign, SO and MM overall. But the AI certainly didn’t help to create many cool / epic memories of battles.

Parts were hard, yes, but it’s not b/c they were ‘playing smart / savvy’. It was just lots of them and legendary difficulty.

Quality over quantity. Bring back the legendary elites that make me want to punt a baby llama.

An example of bad AI is how I can stand partially around a corner and slowly kill a Wraith or Hunter without any movement / shots fired back. The AI often times feels broken.

> But should updating the polygon count or increasing number of buttons in forge have the same priority as the AI?

Arguably, yes. Arguably even more so since people expect a graphical jump since HXB1 is going to be on a next gen console and because H4 was missing so many Forge options.

> My argument is that AI should take precedence over other features, not that it should be the ONLY one developed though I did use that as an extreme example to illustrate the point (it’s difficult to make the case that any other single update would be able to justify a new game. They just don’t have the same potential, hence why the most effort should be allocated to exploiting the huge opportunity that lies in more fundamentally variable AI.)

Why?

I don’t see the majority of player interest isn’t fighting the AI. At least not for extended periods such as the life of the game. It’s in dominating in the MP. It’s in building cool playgrounds. It’s in taking awesome screenshots. It’s in finding secrets that the developers buried in the levels. It’s in the narrative and story.

You want to extensively polish an area that not many people are going to look at.

> And how do you make that happen? By adding +1 to all facets?

~_^

Please don’t assume I’m that naive.

> You need to set priorities based on not only where Halo’s strengths are (so it can continue on having a unique competitive identity) but also where it can be most improved.

Yes: You build on the strengths and shore up the weaknesses.

And H4’s AI was weak. But there many other problems that need work as well too that are arguably just as important if not more so. One of my first moments with H4 was right after you literally kickstart the missile into blowing up the Covy ship.

The explosion and breakaway model was god awful. I remember this on top of the poor AI. I remember many other problems as well.

> Now by all means 343 should do something about the graphics, do something about balance. It’s a big studio with lots of different departments working on a collective project. But among them, the most consideration, at least, should be paid to the AI because unlike such things as multiplayer balance it has the definite and achievable potential to advance this franchise beyond anything else we’ve seen.

Assuming that we get extremely complex and almost life like AI: How much credit do you think Halo is going to get for that? Not much by my take. Heck, FEAR set the bar in terms of AI intelligence but that didn’t do much in the long run for the game. And any credit will be lost if other parts of the game are not up to shore.