Generative AI in games – still complicated

Rainer Etzweiler

19.1.2026

Translation: Elicia Payne

What started out as a tool has become an ideological battlefield. Between outraged fans and indifferent corporations, the question arises as to how we as consumers want to approach AI in games – and whether we even have a choice.

Razer has just unveiled Project Ava, its new AI gadget – a 14-centimetre tall hologram. The company calls it a «gaming companion» that virtually «lives next to you», supports you and learns from your interactions. After a week or so, the Waifu Tamagotchi will probably have white sticky stuff on it, be reciting Reddit memes and asking whether there are still enough cans of Monster Energy in the fridge.

Ava’s already a hot contender for the most unnecessary crappy product of the year, and January isn’t even over yet. But this is just the beginning: AI and gaming will continue to merge this year. In an absurd way, as in the case of Ava, but above all in a far more intense way that will change video game development forever.

With that, I’m taking stock.

The forefathers: procedural level generation

Beneath Apple Manor was released for the Apple II in 1978. The fantasy game’s considered one of the first representatives of the roguelike genre. The dungeons in the game are generated algorithmically, so that each restart brings a different level layout. This is referred to as procedural generation. What was revolutionary back then can be seen everywhere nowadays, in titles such as Spelunky, Hades and Dead Cells.

You could argue that these early random number generators are the forefathers of today’s generative AI. That would be a bit of a bold statement though, and not entirely correct.

Procedural generation follows fixed rules i.e. if condition X is fulfilled, element Y appears. The algorithm only follows instructions. Generative AI, on the other hand, operates on a fundamentally different level – it’s been fed millions of data points and spits out content based on statistical probabilities.

It’s worth repeating this is a lame comparison. Nevertheless, it shows that developers have been producing games with the «creative» support of a machine for a long time. So you can see why, to a certain extent, publishers see an opportunity in generative AI.

Gamers, on the other hand, don’t approve. According to a recently published survey by Quantic Foundry, around 85 per cent of respondents have a negative attitude towards the technology. The reasons being far from simple.

Where does the AI defence reflex come from?

The aversion to generative AI stems from various sources. Most of them are absolutely justified and now widely known. Still, I’d like to reiterate the most important points.

First of all, the data question: what were these models trained with? In the vast majority of cases it’s stolen images, texts and music – material the original creators never gave their permission for, nor were compensated for. Surprisingly, the fact AI providers are now earning billions with a product that’s based on the unpaid work of artists has hardly had any legal consequences.

Energy consumption’s another issue: generative AI is a power guzzler of biblical proportions. The data centres required to operate these models consume more energy than some developing countries. The fact that we’re additionally heating up our planet because a few sickos like to drool over pictures of «Sonic» characters with big titties is somewhat worrying.

Then there are the rationalised jobs and the fact that most gamers don’t want to be served art (we’ll leave the discussion of whether games are art for another day) by a soulless machine.

And then there’s the missionaries – who have no choice but to be against AI.

Supporters

Disruptive technologies have never had it easy. Cars were ridiculed as being «horseless carriages» and the internet was dismissed as a gimmick for nerds. Even completely detached from the points mentioned above, it would be legitimate for AI to trigger a certain amount of scepticism. Especially since these are often formed on the sceptic’s ignorance of technological advances.

Cars and the internet eventually caught on. Firstly, because they were useful inventions that offered real added value and secondly, because the supporters behind them weren’t 90 per cent billionaires and the most insufferable «rise and grind» types. A huge part of AI’s image problem is the people behind it.

It’s the kind of people who flex on LinkedIn that they get into ice baths in the morning and wrestle with black bears before underachievers like me have had their first coffee. It reminds me of the NFT enthusiasts who burst into every conversation a few years ago to explain why the JPEG of an ugly monkey was the future.

I sense the same vibes in many AI evangelists today: the same missionary zeal, the same inability to interpret criticism as anything other than «fear of progress» and the same tendency to label anyone who isn’t immediately enthusiastic as a backward Luddite.

What do developers think?

To the surprise of no one, one of the first known controversies on the subject came from Ubisoft. At the beginning of 2023, the developer presented his «Ghostwriter», an AI tool designed to generate dialogue options for NPCs.

No one knows whether the technology will be used in current games. However, judging by the moral compass of the gaming company, I’d bet a lot of money that all recent releases contain AI babble.

EA CEO Andrew Wilson – whose desire to become a real boy remains unfulfilled – was even clearer to Pocketgamer: «AI is at the very core of our business.»

The crux of the AI debate, however, is that it’s not just the usual suspects – faceless mega-publishers with the conscience of a seal slaughterer – that have come in for criticism. Even studios actually considered «the good guys» have to face uncomfortable questions.

In December 2025, Swen Vincke, CEO of Larian Studios (Baldur’s Gate 3), confirmed that his studio uses generative AI. They use it in PowerPoint presentations, early concept art and placeholder text, as he explained to Bloomberg. The fans’ reactions were, well, far from ideal.

After the interview was published, many gamers vented their resentment on social media. Some declared Baldur’s Gate 3 retroactively worthless, others called the game «slop» and quite a few even attacked Vincke directly.

The reactions were fierce, especially when you consider the currently very harsh tone that prevails online.

Vincke promptly rowed back and clarified on X: «Holy fuck we’re not (…) replacing concept artists with AI.» The studio employs 23 concept artists and no AI-generated content makes it into the finished games.

Sandfall Interactive, the studio behind Clair Obscur: Expedition 33 also had a similar experience. The RPG was one of the biggest surprise successes of recent years. Fans were ecstatic, and so were the critics, which was reflected in numerous awards for the Frenchman’s debut.

However, two of these Indie Game Awards, were recently withdrawn from the game when it was discovered that AI-generated placeholder textures had been used in the early development phase.

It’s hard to see how being this pedantic is the right way to go.

Is there an ethically correct way to use it?

I’m pretty sure most big developers don’t give a damn whether their use of technology is done in a responsible way. Still, I hope that a few decision-makers will look into the question of whether generative AI can be used sensibly in video game development.

Theoretically, there are quite a few options on the table. Perhaps when the tools only have access to their own files or take on pointless stuff that requires neither artistic talent nor a creative approach.

In practical terms, however, I’m less optimistic. Listed companies prioritise quarterly figures, not ethical principles. If AI lowers production costs, it will be used – regardless of who gets the short end of the stick. The same companies would probably also send orphans to hydrochloric acid mines if it didn’t happen to be against the law.

There’s no going back

Fact is, we have to look at the bad parts: the Pandora’s box has been opened, the genie’s out of the bottle.

Generative AI won’t disappear from the gaming industry. The technology’s too powerful and too tempting for companies looking for efficiency gains. The indignant – and I count myself among them – are, in the grand scheme, a small fringe group. The overwhelming majority of consumers are casual gamers who aren’t deeply involved in the subject matter and barely notice the controversies.

One option for a silent protest would be voting with your wallet i.e. share your opinion by only buying the products you like. But let’s be honest: people are incredibly inconsistent. I’ve known for years that Spotify’s a terrible platform that exploits musicians and whose founder invests in warmongering. And yet, have I changed? No. I was too lazy and still am.

The conclusion is therefore somewhat sobering: AI in games is here to stay – like microtransactions, season passes and «Definitive Editions». Whether we like it or not is of secondary importance to the decision-makers. The only question that remains is how much of our collective protest are we gamers prepared to turn into actual action?

History would say: not very much.

Rainer Etzweiler

Freelance Editor

rainer@rockstar.ch

In the early 90s, my older brother gave me his NES with The Legend of Zelda on it. It was the start of an obsession that continues to this day.

Follow

---

Gaming

Follow topics and stay updated on your areas of interest

Follow topic

---

Opinion

This is a subjective opinion of the editorial team. It doesn't necessarily reflect the position of the company.

Show all

These articles might also interest you