Exactly, as I don't expect QA done by something that can't think or feel to know what actually needs to be fixed. AI is a hallucination engine that just agrees rather than points out issues, in some cases it might call attention to non-issues and let critical bugs slip by. The ethical issues are still significant and play into the reason why I would refuse to buy any more Square Enix games going forward. I don't trust them to walk this back, they are high on the AI lie. Human made games with humans handling the QA are the only games that I want.
Exactly, as I don’t expect QA done by something that can’t think or feel to know what actually needs to be fixed
That is a very small part of QA’s responsibility. Mostly it is about testing and identifying bugs that get triaged by management. The person running the tests is NOT responsible for deciding what can and can’t ship.
And, in that regard… this is actually a REALLY good use of “AI” (not so much generative). Imagine something like the old “A star algorithm plays mario” where it is about finding different paths to accomplish the same goal (e.g. a quest) and immediately having a lot of exactly what steps led to the anomaly for the purposes of building a reproducer.
Which actually DOES feel like a really good use case… at the cost of massive computational costs (so… “AI”).
That said: it also has all of the usual labor implications. But from a purely technical “make the best games” standpoint? Managers overseeing a rack that is running through the games 24/7 for bugs that they can then review and prioritize seems like a REALLY good move.
They're already not paying for QA, so if anything this would be a net increase in resources allocated just to bring the machines onboard to do the task
I would initially tap the breaks on this, if for no other reason than “AI doing Q&A” reads more like corporate buzzwords than material policy. Big software developers should already have much of their Q&A automated, at least at the base layer. Further automating Q&A is generally a better business practice, as it helps catch more bugs in the Dev/Test cycle sooner.
Then consider that Q&A work by end users is historically a miserable and soul-sucking job. Converting those roles to debuggers and active devs does a lot for both the business and the workforce. When compared to “AI is doing the art” this is night-and-day, the very definition of the “Getting rid of the jobs people hate so they can do the work they love” that AI was supposed to deliver.
Finally, I’m forced to drag out the old “95% of AI implementations fail” statistic. Far more worried that they’re going to implement a model that costs a fortune and delivers mediocre results than that they’ll implement an AI driven round of end-user testing.
Turning Q&A over to the Roomba AI to find corners of the setting that snag the user would be Gud Aktuly.
And hire other people with the excess budget. Hell, depending on how badly these systems are implemented, you can end up with more staff supporting the testing system than you had doing the testing.
If they can do their work more efficiently, they don’t get laid off.
It just means a better % of edge cases can get covered, even if you made QAs operate at 100x efficiency, they’d still have edge cases not getting covered.
I was going to say, this is one job that actually makes sense to automate. I don’t know any QA testers personally, but I’ve heard plenty of accounts of them absolutely hating their jobs and getting laid off after the time crunch anyway.
They already have a really cool solution for that, which they talked about in their GDC talk.. I don’t think there’s any need to slap a glorified chatbot into this, it already seems to work well and have just the right amount of human input to be reliable, while also leaving the “testcase replay gruntwork” to a script instead of a human.
Take-Two’s CEO doesn’t think a Grand Theft Auto built with AI would be very good | VGC
Sounds fair to me, at least for near-term AI. A lot of the stuff that I think GTA does well doesn’t map all that well to what we can do very well with generative AI today (and that’s true for a lot of genres).
He added: “Anything that involves backward-looking data compute and LLMs, AI is really good for, and that and that applies to lots of things that we do at Take-Two. Anything that isn’t attached to that, it’s going to be really, really bad at…. there is no creativity that can exist, by definition, in any AI model, because it is data driven.”
To make a statement about any AI seems overly strong. This feels a little like a reformed “can machines think?” question. The human mind is also data-driven; we learn about the world, then create new content based on that. We have more sophisticated mechanisms for synthesizing new data from our memories than present LLMs do. But I’m not sure that those mechanisms need be all that much more complicated, or that one really requires human-level synthesizing ability to be able to create pretty compelling content.
I certainly think that the simple techniques that existing generative AI uses, where you just have a plain-Jane LLM, may very well be limiting in some substantial ways, but I don’t think that holds up in the longer term, and I think that it may not take a lot of sophistication being added to permit a lot of functionality.
I also haven’t been closely following use of AI in video games, but I think that there are some games that do effectively make use of generative AI now. A big one for me is use of diffusion models for dynamic generation of illustration. I like a lot of text-based games — maybe interactive fiction or the kind of text-based choose-your-own-adventure games that Choice of Games publishes. These usually have few or no illustrations. They’re often “long tail” games, made with small budgets by a small team for a niche audience at low cost. The ability to inexpensively illustrate games would be damned useful — and my impression is that some of the Choice Of games crowd have made use of that. With local computation capability, the ability to do so dynamically would be even more useful. The generation doesn’t need to run in real time, and a single illustration might be useful for some time, but could help add atmosphere to the game.
There have been modified versions of (note: very much NSFW and covers a considerable amount of hard kink material, inclusive of stuff like snuff, physical and psychological torture, sex with children and infants, slavery, forced body modification and mutilation, and so forth; you have been warned) that have incorporated this functionality to generate dynamic illustrations based on prompts that the game can procedurally generate running on local diffusion models. As that demonstrates, it is clearly possible from a technical standpoint to do that now, has been for quite some months, and I suspect that it would not be hard to make that an option with relatively-little development effort for a very wide range of text-oriented games. Just needs standardization, ease of deployment, sharing parallel compute resources among software, and so forth.
As it exists in 2025, SillyTavern used as a role-playing software package is not really a game. Rather, it’s a form of interactive storytelling. It has very limited functionality designed around making LLMs support this sort of thing: dealing with a “group” of characters, permitting a player to manually toggle NPC presence, the creation of “lorebooks”, where tokens showing up trigger insertion of additional content into the game context to permit statically-written information about a fictional world that an LLM does not know about to be incorporated into text generation. But it’s not really a game in any traditional sense of the word. One might create characters that have adversarial goals and attempt to overcome those, but it doesn’t really deal well with creating challenges incredibly well, and the line between the player and a DM is fairly blurred today, because the engine requires hand-holding to work. Context of the past story being fed into an LLM as part of its prompt is not a very efficient way to store world state. Some of this might be addressed via use of more-sophisticated AIs that retain far more world state and in a more-efficient-to-process form.
But I am pretty convinced that with a little work even with existing LLMs, it’d be possible to make a whole genre of games that do effectively store world state, where the LLM interacts with a more-conventionally-programmed game world with state that is managed as it has been by more traditional software. For example, I strongly suspect that it would be possible to glue even an existing LLM to something like a MUD world. That might be via use of LoRAs or MoEs, or to have additional “tiny” LLMs. That permits complex characters to add content within a game world with rules defined in the traditional sense. I think I’ve seen one or two early stabs at this, but while I haven’t been watching closely, it doesn’t seem to have real, killer-app examples…yet. But I don’t think that we really need any new technologies to do this, just game developers to pound on this.
Personally I think AI generated content could be great when it’s used to create content that otherwise wouldn’t be present. Like when you have a game where all the buildings are just static models with all the doors closed and the curtains shut, imagine resolving all that with buildings you could go in. Basically I want Cyberpunk where all the lights and movement actually mean something.
Basically I want Cyberpunk where all the lights and movement actually mean something.
totally valid desire, but I don’t think AI would give you that solution. If you went into a building and it was a weird, hallucinated backroom, would that give you that feeling that you’re looking for? Or would you be left feeling disappointed in a different way?
People often don’t realize that most things can be and have been done with very simple algorithms, more advanced algorithms or at most very simple neural networks. Instead, they immediately jump to LLM integrations.
Training a model to generate 3D models for different levels of detail might be possible, if there are enough examples of games with human-created different-LOD models. Like, it could be a way to assess, from a psychovisual standpoint, what elements are “important” based on their geometry or color/texture properties.
We have 3D engines that can use variable-LOD models if they’re there…but they require effort from human modelers to make good ones today. Tweaking that is kinda drudge work, but you want to do it if you want open-world environments with high-resolution models up close.
I ultimately don’t think he’s a bad guy so I’m rooting for him to show us wrong.
I think he always just believed in his vision too much and didn’t realise that he and his team couldn’t end up delivering on it. And then he’d be so excited about what he’s working on that he just couldn’t not share it with others. I don’t think he ever intended to mislead.
I know there’s the milo thing but with that I feel like he still thought they could do it in the end but Microsoft just needed something right there and then to show. I think that issue ends up more squarely on whoever sold him on what the hardware could do and he had not yet fully hit those walls of limitations enough to realise they weren’t coming down at all. (I also think this goes for Sean Murray too when he did the talk show circuit that Sony put him on - although without the hardware stuff and more on whether or not they could deliver on time since they’ve shown what they wanted was actually possible with future updates)
I’m taking the bait again for old times sake if nothing else. I hope he proves everyone wrong, it will be a fantastic end to his saga, and we’ll get something great.
When we’re let down again, we’ll still have something fun to play with and getting tricked by Molyneuxs big ideas will be like opening a time capsule from a more civilised age.
Peter Molyneux says Masters of Albion will make up for decades of ‘overpromising on things’
When you “make up” for something you’ve done, it’s not up to you to decide when you’ve “made up” for it. It’s up to the people that you crossed. What a presumptuous thing to say.
The funniest thing he ever tried to sell was Milo.
Basically an AI character game before even shitty LLM generative AI existed. You had to have been the dumbest, most gullible rube to have believed anything about that shit.
It’s obvious that demo was 99% fake and the eventual end product would have been way more scripted and simple than what he hyped it to be (I mean, it’s Molyneux). But he’s also been backstabbed by Microsoft on that one.
The Kinect prototype he was working with was not the Kinect that was eventually released. At one point Microsoft cut corners and removed the internal processor that was suposed to make Kinect work, leaving the console to deal with all the extra computation. It was barely possible to make a simple Kinect game not run like shit, making something relatively smart and responsive would have been a pipe dream.
I was fairly young at the time and wondered how they would try to pull this off. To the surprise of no one who was a little older than me they simply didn‘t. The only thing that was pulled was the project itself.
It’s been awhile since I played Godus, but I don’t remember micro transactions. May have forgotten, but remember it just being a super simplified version of populous. I only played in early access though, not sure on changes or if it ever had a full release.
videogameschronicle.com
Gorące