In the upcoming issue of Game Informer magazine (subscribe here), we have a feature about the finale of the Talos Principle trilogy, The Talos Principle III. We got feedback from creative director Davor Hunski about what to expect in the final entry in the series, but we also spoke with writers and narrative designers Jonas Kyratzes and Verena Kyratzes, who had many thoughts on AI, both the fictional general intelligence kind and the modern generative AI kind.
Since the original Talos Principle in 2014, the series has thoughtfully considered the nature of what it means to have consciousness, to be a person, and how a genuine, sentient artificial intelligence might exist in the long term. Considering the two have been thinking about AI longer than most, I asked them for their thoughts. You can check out the interview excerpt below.
Do you feel like the rest of the world is almost late to the party in terms of thinking about AI and what it means to humanity? This is something you all have been percolating on for years now.
Jonas Kyratzes: It’s so complicated because the current AI conversation is kind of a bit strange in that some of the time we’re talking about something that we call AI, but it’s not artificial general intelligence. It’s LLMs or machine learning in various forms, which aren’t necessarily the same as artificial intelligence in the sense of conscious being. And Talos Principle…well, it’s not really about AI, in the sense that we use artificial intelligence as a way to talk about materialism. So there is, obviously, a relevance to AI in the sense that if we manage to create an actually thinking being out of technology, that obviously has vast philosophical implications. And those are the things that the game – the first game, and to some degree the later ones – are particularly concerned with, right? The material nature of consciousness. That’s what’s the most relevant to the game specifically.
The other applications that we are discussing nowadays in our world – which don’t actually have to do with intelligence – are not so relevant directly to the games, except then again when it comes to the question of technology in a bigger sense, where the second game is very concerned with how do we apply technology correctly and what does that mean for us as human beings? And how can we use it to liberate ourselves or destroy ourselves according to various perspectives? So, there is a relevance there in general, of course, to you know any kind of labor-saving technology or technology that expands our horizons, but not specifically to what we are calling AI now in our world. Kind of a complicated answer.
It’s a complicated topic. There have been AI – LLMs genAI – CEOs who claim we’ve hit the singularity, which to me, that seems absurd. But I’m curious what you two think of a statement like that in 2026.
Jonas: I’m going to talk for myself. I think it’s largely ludicrous. I think the technology itself is very impressive, and I think machine learning has tremendous applications, most of which we’re currently not seeing or getting, I think, because in a profit-oriented system, a lot of the things that we needed to do, we’re just not going to use it for, because we’re going to use it for stupid things instead, for people to try and make vast amounts of money before their companies collapse. So, it’s complicated. The technology itself is clearly very, very impressive and has a number of very interesting applications. But at the same time, these excessive quasi-religious claims that are made for it are… they’re just not rooted in any kind of reality. And I think a lot of them are also mainly just meant to juice up the stock market.
Verena Kyratzes: I’m with Jonas in that the LLMs that we’re currently dealing with? They’re basically incredibly intelligent versions of the autocorrect that I have on my phone, and calling them actual AI, general AI, is just ludicrous.
The mission statement of Talos II was to talk about the use of technology, the ethical use of technology, and a bit breaking down this fear of technology that a lot of people have, and in that context, this current AI discussion – and there’s going to be a giant, “but,” at the end of what I’m saying, by the way – but this current AI discussion is very bothersome to me because I like to go into these Wikipedia deep dives and read like in the 1800s, the first passenger trains happened, and everybody was like, “Oh no, my God! It’s going to turn our brains into paste. And then the advent of cinema, and everybody’s like, “No, it’s going to give people psychosis!” And now we’re here, and suddenly everybody is treating AI the same, and I’m with them. And I don’t want to be with those people because I always want to approach every new technology with an open mind, and say, “You know… maybe it is good. Maybe we can use it.” But I think in the current system that we have, it’s just a plagiarism machine that makes people lose their jobs. But I want to have hope that it’s a good thing. It’s very troublesome to me.
I appreciate that answer. Plagiarism machine that destroys jobs is like a great summary. I assume over the course of development that Croteam probably doesn’t really engage with genAI in terms of creativity. Do you feel it hinders creativity?
Verena: Yes.
Jonas: I think when it is used, it hinders it because… okay, I have more complicated feelings about this because, like I said, I have tried it, I’ve used it, I’ve experimented with it to have an idea of what I’m talking about, and I do think that it has applications that are perfectly legitimate, and I will reach for it, for example, when I’m trying to find a rhyme or a particular word that I can’t remember. It works great as a thesaurus, and it has certain applications that are perfectly legitimate. But you have to really know what you’re doing already to use it safely.
It’s a bit like an actual thesaurus in that people who are not very good writers will reach for a thesaurus and then just try to replace every word with a word that has a similar meaning and create this monstrosity of a text. Whereas a thesaurus, of course, is actually for remembering words that you already knew the meaning of, and you’re just like, “What’s that synonym? God damn it!” And then you look it up in a thesaurus, and you find that’s the use of that thing.
In the same way, there are applications, but they take a skilled user, especially when it comes to something generative to even use. Most applications are very dangerous because, of course, as a writer, you need to be writing. And also as a writer, you need to be capable of evaluating the quality of what you’re producing. What we of course see with LLMs is a lot of people who know nothing about writing just being like, “No, no, I can replace all my writers with this.” And you can’t because it sucks.
That said, I’m not against any technology in and of itself. So, to me, a lot of the questions are not questions of the technology. For example, when a lot of jobs are being lost because of AI, I don’t think that that’s an AI issue. I think that’s a labor issue. That is, why is the power of labor relative to capital so weak? So low that we can’t defend ourselves against the owners just eliminating everyone? Like, what’s happened to, you know, militant labor, for example. Where is that?
A lot of the time, the question is not the technology in and of itself, but the social and political context in which the technology is applied. That’s a lot of what Talos II was about. Can we use technology safely? Is the technology inherently, somehow harmful or bad? Or is it the context of society? And can we build a better society where these things are used in a meaningful way? And I would say the same applies to machine learning. If we apply it for the benefit of humankind, then it will benefit us. And if we apply it so that five people can make a trillion dollars, then it will not end well. But we do have the choice of applying it one way or the other.
Verena, you had a very succinct answer when I asked if it hinders creativity.
Verena: It absolutely hinders creativity. What I have encountered in the wild, not at Croteam mind you, but I have encountered this in the wild that people were like, “Hey, I’ve written a rough draft of the text with AI, and then you can just fill in the gaps.” And I’m like, “No, that’s not helpful. Now I have those s*** sentences stuck in my head, and I can’t get them out.” So, it’s very dangerous because even if you just do placeholder text, the artifacts kind of… I mean, they remain. You can’t get rid of all of it, for me as a writer, at least. It’s incredibly hard when I see something. But that applies just as well… like we worked on a game ages ago where just the level designers were filling out temp text, and that’s just the same problem. It’s not specific to AI. It’s just very hard to get rid of those artifacts, so to speak.
And to recenter us, how are these thoughts that are swimming around your head feeding into Talos III? Is Talos III examining this modern version of AI? Or were you all already planning this story before it became such a hot topic?
Jonas: No, Talos III really looks beyond where Talos II ended. To a world of freedom. Freedom from necessity, as it says somewhere. A world where people have built a better world that is no longer driven by the same forces as our world. It does the opposite. It tries to imagine a world that’s much better than ours, in which people had much better lives than ours, and to give us the idea again that it is possible. That the world we live in now is not the only version of things, and it’s a world of abundance and equality and freedom for people.
Artificial intelligence is tied to it in the sense that obviously the characters from the first game were artificially intelligent beings that then continued into the second game, and then into the third, and have now expanded beyond Earth to the stars. But […] the second game has as its premise that the characters consider themselves human; they are robots, but they just consider themselves fully fledged human beings. They are the descendants of humanity. So, I guess you can say that that is an optimistic view of the possibilities of artificial intelligence because it’s not one where they become some sort of giant machine god that rules everything. They just become people, and you even see a number of characters in Talos III that just choose to remain essentially a humanoid shape, live relatively ordinary lives, even though they’re immortal and they live for billions of years. They don’t choose to become alien and strange. They choose to enjoy many of the things that we human beings enjoy because they are human beings, and those things are good things that should be enjoyed.
So, it’s about these artificial intelligences that, for example, in the DLC for Talos II, which is pretty important part of the story, Road to Elysium, you have them developing taste buds so they can taste stuff and just learning pleasure and happiness and finding out how to have children and just basically trying to live. And there’s a recurring idea, which is that they don’t want to be like post-human, but they want to be more human. They want to just have that which makes us human, more of it, enjoy life more than we could. So, that’s kind of the direction that we are heading in with that story. Now, obviously, that is a kind of commentary on technology and the possibilities of technology, but it’s not a commentary specifically on, “Oh, they used machine learning to do this or that.”
It’s almost coincidental to a certain degree.
Jonas: Right. It’s thematically relevant, but it’s not specifically about that.
For more on The Talos Principle III, check out the next issue of Game Informer magazine. You can subscribe right here.