Greetings, glorious testers!

Check out Alpha Two Announcements here to see the latest news on Alpha Two.
Check out general Announcements here to see the latest news on Ashes of Creation & Intrepid Studios.

To get the quickest updates regarding Alpha Two, connect your Discord and Intrepid accounts here.

AI! Is there a race against time?

Individuated SoulIndividuated Soul Member, Alpha One, Alpha Two, Early Alpha Two
I know AI has been brought up a few times for NPCs, but the current trajectory of AI and the acceleration of progress, it makes me wonder if releases (ones that are ~2 years out say) are in a race against time. The idea is that there will be a point that AI will be able to simulate and iterate virtually exponentially better than then your average game developer. When this holds true, more perfect worlds will be able to be created at a fraction of the development cost. I know this might sound like science fiction, but the train is arriving to the station faster than anyone has initially predicted. The picture is from NVIDIAs release of Blackwell, and it should help paint the picture.

Should ashes account for this? Is this nothing but a positive and there is nothing to worry about? Is Ashes planning to use AI? I would be worried to hear that AI isn’t being seriously looked at considering it appears that they have a long way to go.

Just a discussion to be had.


o0xhkftyyuc1.png
xCSOHOG.png
«1

Comments

  • AzheraeAzherae Member, Alpha One, Alpha Two, Early Alpha Two
    edited April 11
    No, that's basically not how that works.

    AI generally will create either 'consensus', or 'constrained' things, and is currently (and is not soon likely to change) not particularly good at anything inbetween.

    So it's nothing but a positive, generally.

    All it means is that less skilled designers will put out more products and dilute the space.
    ♪ One Gummy Fish, two Gummy Fish, Red Gummy Fish, Blue Gummy Fish
  • Individuated SoulIndividuated Soul Member, Alpha One, Alpha Two, Early Alpha Two
    edited April 11
    Which part you are referring to when you said "thats basically not how that works"? May you also explain 'consensus' or 'constrained' in AI context? Also the question is not currently, but a near future state (relatively speaking). AI is currently being trained via artificial realities as the internet is not enough apparently for AGI.

    I think the AI will be the developer in this case, not humans.

    Example where an AI interfaces with other agents currently (from the same conference):

    247ycvpxby2o.png

    Let me say that I am no way an expert on this. Just bringing this up for discussion ^^;
    xCSOHOG.png
  • AzheraeAzherae Member, Alpha One, Alpha Two, Early Alpha Two
    The part I was referring to is the 'race against time' aspect, basically.

    No one is in a race with the AI, the AI is generally told 'which data to look at' and 'what the prompter needs it to do or interpret from that data'.

    Then, the AI will either make a 'consensus' plan or product, in the style of:
    "This is what 90% or more of people/experts say is the answer."

    Or it will make a 'constrained' plan or product:
    "Here is what you asked me to make, the best fit for all the parameters you provided."

    The first type would never be 'original' because everyone would ask basically the same question and get the same answer.

    The second type might not be 'good' because the asker doesn't know the required parameters, and if the parameters are complicated or uncommon, the AI needs a lot of data and a lot of tokens to achieve the goal.

    AI is best at 'extrapolating from the collective knowledge of humanity, especially when they agree on a solution'. So, I guess I'm saying that for AI 'design' to create good games, we'd have to agree on what those are, and in many cases we'd have to have made them already.
    ♪ One Gummy Fish, two Gummy Fish, Red Gummy Fish, Blue Gummy Fish
  • Individuated SoulIndividuated Soul Member, Alpha One, Alpha Two, Early Alpha Two
    Understood, but you keep describing current systems today (predictive and tasked based). Elon Musk predicts AGI as early as 2 years (possibly but 2027-29 higher probabilities).

    Here is another picture (Source:https://viso.ai/deep-learning/artificial-intelligence-types/):
    j6n0axe5tqzv.png


    xCSOHOG.png
  • AzheraeAzherae Member, Alpha One, Alpha Two, Early Alpha Two
    Understood, but you keep describing current systems today (predictive and tasked based). Elon Musk predicts AGI as early as 2 years (possibly but 2027-29 higher probabilities).

    Here is another picture (Source:https://viso.ai/deep-learning/artificial-intelligence-types/):
    j6n0axe5tqzv.png


    Without disparaging the Musk, even if he is right, being 'smarter than humans' doesn't really translate to 'can build an MMO' because the problem space isn't necessarily that type.

    If you're referring to the fact that a really strong designer could build massive games without needing to hire average developers, I can see that happening, but I'm not sure who is 'racing' against the AI in that case.

    Please note that I have a very biased viewpoint here because I'm also a data engineer with some basic projects who works with an artist, network 'consultant', and have a prompt engineer. So maybe the only reason I don't see the 'problem', is that I'm part of it.
    ♪ One Gummy Fish, two Gummy Fish, Red Gummy Fish, Blue Gummy Fish
  • CROW3CROW3 Member, Alpha Two
    The most accurate response to this is: maybe.

    It’s hard to predict the extent that AI will impact game development (and what aspects of development). For as interesting as GenAI is, and how much attention is garnered, it’s wholly constrained by the bounds of thoughts that have been already been thought and documented (LLMs). AI is good at quick synthesis and generally organizing that information, but this hasn’t really been unleashed to push beyond defined limits.

    I don’t think we’re going to see games cut from while AI cloth for some time, but I imagine we’ll see some good acceleration here and there to specific aspects in the near future.
    AoC+Dwarf+750v3.png
  • nanfoodlenanfoodle Member, Founder, Kickstarter, Alpha Two, Early Alpha Two
    edited April 12
    AI has gone from the IQ and wisdom of a average teenager to an IQ of about 150 in about 5 years. Experts are saying that in the next few years AI will have the IQ of about 5000. It's wisdom level? Fact is, there will be very few areas of life that won't be impacted by AI in a large way.

    AI can now make 3D spaces in a few min just from a 2D piece of art it give it the concept of the style it makes it in. It's not that big of a leap to make 3D worlds next.

    How long will it take that these 3D worlds look hand crafted? How long before a small team can make an MMO and use AI to make story, quests, world's, classes all by just giving a discription of what the developer wants to AI? Yes this is all coming. Is this a good thing? Yup allot of humans are going to be put out of work. Where you needed 50-200 people to make a game. Will one day be made by a small handful. Good side of that. How much will games cost when the market is flooded? Again, is this a bad thing or a good thing?
  • DygzDygz Member, Braver of Worlds, Kickstarter, Alpha One, Alpha Two, Early Alpha Two
    We will have flying cars before 2020.
  • nanfoodlenanfoodle Member, Founder, Kickstarter, Alpha Two, Early Alpha Two
    Dygz wrote: »
    We will have flying cars before 2020.

    The first real flying car was made in 1935. Fast forward today and we don't have mainstream flying cars more because of laws then the tech. AI will not have the same constraints as the government, its their next big weapon and the race is on.
  • VyrilVyril Member, Alpha Two
    edited April 15
    If it can be weaponized effectively there is no stopping the tech to improve. Money, time, and energy will be thrown at it.

    That's the true race.
  • blatblat Member
    Be prepared for highly speculative / subjective statements delivered as absolute objective fact.

    Tbh on a subject like this, the more certain someone sounds the more likely they're full of sh1t.
  • CROW3CROW3 Member, Alpha Two
    blat wrote: »
    Tbh on a subject like this, the more certain someone sounds the more likely they're full of sh1t.

    So certain are you? 😆

    AoC+Dwarf+750v3.png
  • KilionKilion Member, Alpha Two
    Even if games in the future could be created by AIs (which I doubt to be something we will see on the scale of an MMO within the next 10 years) who says that HUMANS will like those? We obviously can't even decide if beautiful female character models in games "can be allowed". And if I look at similar developments in the story writing industry we are facing similar issues. And on that we now throw a new technology that generates black WW2 soldiers? I'm not convinced. Especially with all the lawsuits regarding copyright boiling up and energy necessary to run such a thing. And wasn't there this software engineer from Googles AI department who recently said that progress is largely overestimated?

    So IMO there is plenty of time where we will stick with games designed largely or completely by humans.
    The answer is probably >>> HERE <<<
  • blatblat Member
    CROW3 wrote: »
    blat wrote: »
    Tbh on a subject like this, the more certain someone sounds the more likely they're full of sh1t.

    So certain are you? 😆

    Ha alright he's got me ;)
  • GrappLrGrappLr Member, Alpha One, Alpha Two, Early Alpha Two
    I am very excited to play AoC because I believe it'll be an amazing gaming experience.

    Will AoC lose out to better MMORPG games a few years later due to rapid AI improvement? Except for Intrepid, that's a good thing for everyone else. More great games? Bring it on!

    End of the day, Intrepid is making the best game they can with the tools they're given.

    Similarly, if AI advances that quickly, Intrepid could as well use it to upgrade their game in various directions. There's nothing saying they have to be stuck in the stone age as things progress. If anything, we've seen Intrepid be very open to updates, as seen from changing UE4 to UE5.
  • AzheraeAzherae Member, Alpha One, Alpha Two, Early Alpha Two
    If anything, Intrepid would have a massive head start because even a non-Generative AI still needs goal parameters and constraints.

    Intrepid, by the time AI consistently reaches that point, would have those. They'd be hugely ahead of any competition because they'd be at the 'here are our problems and pain points, and our underlying structure, help us fix this' instead of 'hey how much PvP should a good game have, Silvia?'
    ♪ One Gummy Fish, two Gummy Fish, Red Gummy Fish, Blue Gummy Fish
  • nanfoodlenanfoodle Member, Founder, Kickstarter, Alpha Two, Early Alpha Two
    edited April 12
    Azherae wrote: »
    If anything, Intrepid would have a massive head start because even a non-Generative AI still needs goal parameters and constraints.

    Intrepid, by the time AI consistently reaches that point, would have those. They'd be hugely ahead of any competition because they'd be at the 'here are our problems and pain points, and our underlying structure, help us fix this' instead of 'hey how much PvP should a good game have, Silvia?'

    That's a big reason to use UE5. It's a good reason to jump on as a developer. It's future proof. It's up to each developer on how it's used.
  • blatblat Member
    Tbh examples like New World & recent WoW show us that technology & finance aren't necessarily the big difference makers they once were.

    As we're seeing in video content; the tech doesn't count for much without the vision. There'll be competition of course but Intrepid could well find themselves in a v good position.
  • AszkalonAszkalon Member, Alpha Two
    The idea is that there will be a point that AI will be able to simulate and iterate virtually exponentially better than then your average game developer.

    Not wanting to underestimate A.I.,

    but i doubt that v~EEERY much that artificial Intelligence could do a better Job at for Example designing a huge and epic World like the World of Verra - than the respective Developers of Ashes of Creation.

    An A.I. can only guess by it's Programming what Human Video Gamers want in an MMO.


    It can never "listen and implement" the Wishes of a Human Playerbase as good and well as a Human Developer. Sure an A.I. doesn't need to sleep, eat, etc. But there are some things only Humans can do.

    It is the same Reason we can leave things like Hospital Care for ill People not to A.I. Some Jobs and Tasks can just not be taken out of Human Hands. Because nothing non-human can replace a Human Being for Jobs and Tasks where literal Humanity and Human-thinking is required.
    a50whcz343yn.png
    ✓ Occasional Roleplayer
    ✓ Kinda starting to look for a Guild right now. (German)
  • Individuated SoulIndividuated Soul Member, Alpha One, Alpha Two, Early Alpha Two
    Agree to disagree. I understand that it may seem farfetched now, but the human experience, humanity, our essence, whatever you want to call it, will be quantified and virtualized. The idea that it will never be good as a human developer is human thinking, or human narcissim.

    Another point, as Elon puts it, we have a bandwidth problem (and hardware problem with our current upper limits of IQ) that limits our effectiveness. AI will not have this problem any time soon with advancements in hardware not stopping but actually accelerating.
    xCSOHOG.png
  • Jam21Jam21 Member
    AI is already used in the process of making games, and i hope Interpid can use the latest developments in their pipeline to speed up AoC release.

    However, actual usage of AI (like in forms of "intelligent" NPCs or whatever) is very very far. Why? Simple - because it will use too much computing power to keep them operating normally. Will be very expensive to run if ran from server-side, and don't even get me started for what it would be like if the LLMs used for in-game purposes would be client-side (and thus modifable by haxxors).
  • nanfoodlenanfoodle Member, Founder, Kickstarter, Alpha Two, Early Alpha Two
    Aszkalon wrote: »
    The idea is that there will be a point that AI will be able to simulate and iterate virtually exponentially better than then your average game developer.

    Not wanting to underestimate A.I.,

    but i doubt that v~EEERY much that artificial Intelligence could do a better Job at for Example designing a huge and epic World like the World of Verra - than the respective Developers of Ashes of Creation.

    An A.I. can only guess by it's Programming what Human Video Gamers want in an MMO.


    It can never "listen and implement" the Wishes of a Human Playerbase as good and well as a Human Developer. Sure an A.I. doesn't need to sleep, eat, etc. But there are some things only Humans can do.

    It is the same Reason we can leave things like Hospital Care for ill People not to A.I. Some Jobs and Tasks can just not be taken out of Human Hands. Because nothing non-human can replace a Human Being for Jobs and Tasks where literal Humanity and Human-thinking is required.

    AI can play millions of game hours on a game in a matter of a few days. You can let AI play a game like Zelda Tears of the Kingdom and in a few months the AI will know every inch of the game and every outcome. Something no human could do in their life time. Then give that same AI Elden Ring and do the same thing. Now ask it to make a zelda like game with Elden Rings art style. Humans are still part of the creative side. But a few days, a week later you have that game. That's not where we are now but that's in the scope of how AI is going and the current direction. This is not fantasy.
  • SathragoSathrago Member, Alpha Two
    Azherae wrote: »
    If anything, Intrepid would have a massive head start because even a non-Generative AI still needs goal parameters and constraints.

    Intrepid, by the time AI consistently reaches that point, would have those. They'd be hugely ahead of any competition because they'd be at the 'here are our problems and pain points, and our underlying structure, help us fix this' instead of 'hey how much PvP should a good game have, Silvia?'

    We have seen examples of them being very smart with how they use their tools. Remember the showcase with how they built an in-editor generator/editor for their assets, letting them skip the blender process?
    8vf24h7y7lio.jpg
    Commissioned at https://fiverr.com/ravenjuu
  • Neox365Neox365 Member
    AI can´t even write Proper C++ code
  • AszkalonAszkalon Member, Alpha Two
    nanfoodle wrote: »
    AI can play millions of game hours on a game in a matter of a few days. You can let AI play a game like Zelda Tears of the Kingdom and in a few months the AI will know every inch of the game and every outcome. Something no human could do in their life time. Then give that same AI Elden Ring and do the same thing. Now ask it to make a zelda like game with Elden Rings art style. Humans are still part of the creative side. But a few days, a week later you have that game. That's not where we are now but that's in the scope of how AI is going and the current direction. This is not fantasy.

    As amazing as this is - why should an A.I. be a better Game Developer, though ? 🤔
    a50whcz343yn.png
    ✓ Occasional Roleplayer
    ✓ Kinda starting to look for a Guild right now. (German)
  • nanfoodlenanfoodle Member, Founder, Kickstarter, Alpha Two, Early Alpha Two
    Aszkalon wrote: »
    nanfoodle wrote: »
    AI can play millions of game hours on a game in a matter of a few days. You can let AI play a game like Zelda Tears of the Kingdom and in a few months the AI will know every inch of the game and every outcome. Something no human could do in their life time. Then give that same AI Elden Ring and do the same thing. Now ask it to make a zelda like game with Elden Rings art style. Humans are still part of the creative side. But a few days, a week later you have that game. That's not where we are now but that's in the scope of how AI is going and the current direction. This is not fantasy.

    As amazing as this is - why should an A.I. be a better Game Developer, though ? 🤔

    That's yet to be determined when it case to quality. Speed is the big reason developers are interested and needing less man power. My worry is games will feel just off enough that games will be in that uncanny valley. Like everything is just a lillte bit off.
  • RocketFarmerRocketFarmer Member, Alpha Two
    I think AI is a tool to be used for a variety of uses.

    The number one utility is time.

    Can be used to reduce the time required to code as well as iteration.

    Has the capability to simulate results for more options/choices in design optimization that could save time in that part of the design.

    Has capability to help in the art and story departments. I still think human guidance is needed for the overall vision of the project. Could make these departments extremely more productive.

    I think the ultimate goal of AI in role playing games is to simulate player actions and behaviors, therefore making the NPCs much more interactive. That can also be utilized to fill out lower population servers to avoid degrading player experience in that situation. Of course the primary limits will be the costs of the infrastructure to accommodate that much AI applicability. Current game AI is used in this fashion, but significantly less powerful that what folks are predicting or hoping for.

    Also need to remember that the idea of something isn’t always realized in the end or in timeframes people anticipate. In the 80s and 90s VR was viewed as the future and dangerous in that people would be lost in that “reality”, however that prediction has only been partially true. You could say social media and smartphones have achieved that far more than VR.

    Gaming is a time sink, but I don’t think AI changes that too much. The MMORPG market is kind of at a low point compared to the past. AI might be able to have the genre see a bounce-back, but I think the AI implementation will be a factor. A poor to cringeworthy AI implementation could cause many to not consume those games. So there’s the potential of a huge risk even if you can save on development costs. But I think it time it will get there, just when is anyone’s guess.

    On the flip side, AI has the potential of affecting the gaming market with a flood of games. Like Kickstarter and Indy games did, but possibly on a wider scale. Could see some very polished games, but play like crap. May depend on the AI utilized to plan, develop and test said game, and what kind of human input is involve. Or in other words, thousands of self published games that you have to buy and play before you can tell whether they are crap or good.

    Suppose AI would also be used for reviews and marketing, so current methods of seeing whether you should buy one may become diluted. That means the #1 currency between developers and consumers will be trust. It really has always been that, but I think AI will enhance its importance. There’s also legal questions regarding AI that we just don’t know at this time. If AI engages in fraud or outright harm, who does the victim seek recovery or compensation? Since those legal questions extend beyond the game, the laws in various regions will have an influence on what can or cannot be done with AI or whether a developer would even want that degree of liability. So the costs of developing a game may just move from your technical staff to your legal team. And I don’t think anyone would consider that progress.
  • CROW3CROW3 Member, Alpha Two
    Hm… can AI write a game so complex that AI can never beat it?

    That’s how you cheese Skynet. Maybe I’ll buy some more Nvidia shares before someone puts that on a product roadmap. 😆
    AoC+Dwarf+750v3.png
  • SathragoSathrago Member, Alpha Two
    Neox365 wrote: »
    AI can´t even write Proper C++ code

    from what I hear its pretty good at doing a bunch of the busy work. If you know coding you just have to correct it from time to time. Just because its not perfectly doing everything for you doesnt make it useless.
    8vf24h7y7lio.jpg
    Commissioned at https://fiverr.com/ravenjuu
  • SunboySunboy Member
    I love the technology and use it everyday. It has flaws like every tool. Many will be ironed out with time. But it still a tool and a tool is only as good as the user.

    Much love <3
Sign In or Register to comment.