This is a really good interview. tl;dw is…
- their next game was going to be D&D, but they changed course and are doing something else now
- Vincke has a vision for “the one RPG to rule them all”, and each of their past three RPGs is a step closer to it
- the next game is not going to be that master vision but one step closer toward it, with their previous 3 RPGs proving out emergent design/multiplayer, story and consequence, and personal stories/performance capture, respectively
- Vincke would like to have this next game done in 3 years compared to BG3’s 6 year development cycle, but realistically expects 4 years, as long as there isn’t something like COVID-19 or a war in Ukraine to impede their progress
I know Lemmy hates it, but I really think AI could play a core part in the future of multiple choice RPGs.
I’m not saying let it be free and build the entire game, but if you train a model to be a certain character and add limitations so it doesn’t go too wild, then that could be massive imo.
Still have a human storyline and imagination behind the game, but use AI like the tool it is for certain parts.
Seriously? You play a game like Baldur’s Gate 3 and your first thought was “damn, this game could really benefit from having less handcrafted, professionally written dialogue”
I speak to enough dipshits at work spewing word salad, this is what I wanted with my escapism, people who follow the fucking conversation not some AI bot resume filling buzzwords about the plot.
In the future I think it’s a really viable option to create more immersive and interactive games. The technology is pretty far away though, not to mention I don’t think most machines could handle the load while also running a game. It’s at best a dream right now, but a pretty interesting idea for 15 years from now.
That’s a pretty big assumption about where the tech is going. In my experience it’s really stupid to try to predict what tech will look like more than a year or two into the future, let alone over a decade.
It can have both.
You could have a fully man made storyline, but then expand the world in a way that is currently impossible.
Even if you train a model for main characters/stories, it would still be built off the work of writers, the model would simply be the character they’ve written.
I think the way video game devs/people are (from what I can see from outside) they are well poised to realize someone making an LLM or a finetune or whatever you want to call it – that produces master level dialogue/stories/whatever is (will be) a skill just like storytelling/writing is.
If I were a JRR Tolkien or Herbert with a universe in my mind, it would be so much more pleasing to make an engine that generates anything from that world that to just write out a few stories from it.
Sounds cool to me
Tolkien was a linguist with a deep fondness for nature and spirituality. He loved creating languages and building beautiful, natural worlds around them. I can’t imagine a single person who would be less enamored by the idea of machinistic language devices that people use to “generate everything”. I think he would be either bored by this possibility or deeply disturbed.
Tolkien also had a deep disdain for industrialism and automation, which is what inspired Isengard in the books. When he says Saruman has “a mind of metal and wheels”, it’s implied that the reader understands why this is a way of saying that Saruman is evil. He definitely wouldn’t be a fan of MindOfMetalAndWheelsGPT.
Perhaps but I can’t see anyone who is interested in creating and communicating fake worlds eschewing the idea too much. If you make a fake world, there’s no way you could ever ‘get it all out’ since you’re just one guy. This would open up that possibility to make a world bigger than yourself and what you can get out of your brain
E: Here’s a long worldbuilding thread about it – https://www.reddit.com/r/worldbuilding/comments/186cspn/your_thoughts_on_the_use_of_ai_for_worldbuilding/
Seems mixed.
One of the foundational tenets of good writing is that worldbuilding is just masturbatory unless it serves the story. You don’t create a cool world and work your way backward into a story. You create a great story and craft a world around it which supports the story you’re trying to tell. The stories are the thing that have value, not the setting or the lore.
Telling a great story is a completely orthogonal skill to worldbuilding, and it requires creativity, emotion, and authorial intent. Star Wars and Harry Potter are both dogshit at worldbuilding, but they’re both some pretty rad stories. Avatar: the Legend of Korra is set in one of the best fantasy worlds ever created and it was a very mediocre story.
Given that is the opposite of what Tolkien did i think you are overstating your case to say it’s a foundational tenet.
Not the opposite at all. Tolkien didn’t know what the One Ring was when he wrote about Bilbo finding it in the Hobbit. Good worldbuilding is iterative. Tolkien went way too obsessive for LOTR and a lot of the worldbuilding he did was purely for his own pleasure rather than serving the story.
Keep in mind he didn’t try to publish The Silmarillion while he was alive. And also that the vast majority of LOTR fans don’t give a shit about stuff in the Silmarillion if it isn’t also relevant to the story of LOTR.
Tolkien spent years creating a fictional world and languages before even deciding to write a novel.
Yeah and my point is that all his worldbuilding was just for his own fun until he actually put in the work of making a story out of it.
Strange, because that is the opposite of every D&D game ever.
The story gets written at the table, at which point the world building should have already been mostly created.
I’m a DM, and I can tell you that as fun as worldbuilding is, no information about your world is real until players learn and remember it. And if you try to loredump on them, they won’t actually remember stuff.
Worldbuilding is fun, but it’s also masturbatory; it’s only fun for the DM until the game’s story makes it matter for everyone else.
I agree with everything you said.
However, fiction world building and game world building are hugely different.
Are games not fiction?
I should have said “literature”.
You only get one BG3 every lifetime though – It was how I thought games would be when I was a kid almost perfectly, but it is the only one…
That statement in itself is quite sad, when one of the reasons everyone called it out as being an amazing game is because it was huge, well crafted, and made by a company that actually seemed to give a shit.
I don’t say this to diminish their achievements, because I’m 80 hours in and still not done, but it’s a spectacularly low bar that Larian absolutely launched themselves over. At a time where companies seem to be scraping the bottom of the barrel, Larian did the exact opposite, and reaped what should be the most obvious of awards (do good work, get lots of money).
I agree – that’s why I never saw the games I dreamed of – Greed. Not some lack of skill, ideas or ability to execute
All came out in my lifetime and my lifetime isn’t even halfway over.
None of those hit for me like BG3 – It’s the perfect CRPG game.
I would have said the same with Divinity 1&2 before BG3 came out
Games keep improving and we will get an even better CRPG in a couple of years that is “even more perfect” than BG3
Maybe now, but everyone was afraid to fund 100MM BG3 because they thought no one wanted CRPG and they want Fortnite or whatever. Hopefully Larian spawns more people who actually care about their users instead of just greed
Frankly I think that’s just recency bias. It’s new so it feels better. Before BG3 came out, most people agreed DA:O was the perfect CRPG, or Mass Effect, and just look at the sheer number of video essays on YouTube praising the quality of F:NV.
New games come along and old games look paltry in comparison. It doesn’t mean the older ones are actually worse. But you’ve had decades to enjoy DA:O, while BG3 still feels like it has secrets to uncover. It still is unexplored territory, and that’s exciting.
Personally I think that once the dust settles, it will be clear that, apart from limitations due to when each was made, these games are all equally 10/10 games in their own way. It’s not as though BG3 is without flaws. And it’s still actively being worked on.
But not me – None of them have ever reached my bar except for BG3 – I care less about video game stories and more about mechanics/freedom of choice to be fair
Chrono Trigger FF3 SNES (FF6) Final Fantasy Tactics FF7
I don’t know what that guy was talking about. There have been so many games released with amazing stories. It’s just the ghost of Jack Welch has slithered into gaming and rather than make great titles we get microtransaction shit games.
Not less handcrafted, but AI enhanced on top of the already excellent written dialogue.
If I want my entire BG3 gameplay to be about grilled cheese, then I would be able to when talking to every NPC while still getting the excellent story about mindflayers. The cheese is just on top.
Unless I see major advancements in the technology, I think AI will be a great tool in the toolbelt for developers operating on lower budgets, but Baldur’s Gate 3 is going to have people expecting the best from Larian’s games going forward, and that’s going to mean human writing and human performances. I think without those major advancements in the technology, it’s going to come off as lesser quality.
I agree, I just think those major advancements will happen.
AI isn’t going to be the answer for everything, but I think it will have a massive impact on video games, in both positive and negative ways.
No doubt some companies are going to be putting out absolute dogshit AI games.
I have used AI to RP some stuff (don’t ask), and while the higher end models, and even the better self hosted models are really good at answering in a way that makes sense and works in context, it is pretty hard to make them do anything, new, interesting, or unexpected, without prompting it specifically.
Nothing that I’ve seen playing around with LLMs makes me think that a well-written work of fiction could be improved by including them, unless there is a significant leap in capability.
And this is ignoring all the discussion about LLMs and copyright/stolen content.
I think AI Dungeon proved that you could have a great Gen AI-driven campaign experience, but the novelty wore out really fast after it was used up as streamer bait and the ethical considerations are just too much of a risk.
If nobody could be bothered to make the thing, I can’t be bothered to play with the thing.
100% the future. Like has anyone put all the billion pages of lore from TES into a GPT finetune? Surely that would make better dialogue than HALT! for the 900000000 time right?