Post Gamescom (Dev) Musings | 02/09/26

Hello all and welcome to this week’s edition of AI and Games. I am back in the dull, grey, and rather wet corner of England I call home and I am thrilled about it. I’m not travelling for another two weeks, I get to sleep in my own bed, the weather is wet, my diet no longer consists largely of sausage and beer - it’s bliss! 14 year old me really hates me right about now.

Last week the broader AI and Games and AI and Games Conference teams had a fantastic time at Gamescom (Dev). It was very rewarding, engaging, and exhausting! I want to take a moment to thank everyone who I chatted with throughout the week for their time, energy, and enthusiasm. Be it our speakers on the AI tracks, our attendees, my colleagues who I shared the stage with in panels and roundtables, and of course to everyone else we had the pleasure of chatting with over a beverage or 12 (those Kölsch really creep up on you). I had so many great conversations throughout the week I’m still trying to process and make notes from it all, so if you’re awaiting a post-Gamescom message please bear with me!

For this week, I figured let’s quickly summarise some thoughts I had from my time at Gamescom. Plus you can expect further Gamescom chat later this week in the AI and Games Podcast as myself and some of our team jump on to discuss the week that was. But before that, the news!

Follow AI and Games on: BlueSky | YouTube | LinkedIn | TikTok

AI (and Games) in the News

Geralt of Rivia: Baby Killer - My favourite story of the past week or so was in a pre-Witcher 3 Remastered world. As reported from an interview in Edge magazine, quest designer Philipp Weber mentioned that an NPC behaviour had to be removed from the popular RPG given it led to a rather awkward scenario. When Geralt appeared in situations that startled NPCs, it would interrupt an activity and should they be holding an item they would also drop it. This lead to the awkward situation where NPCs that were cradling children were dropping babies, which regardless of the game’s mature themes isn’t really something you want to normalise!

ScarJo and Eminem’s Lovechild in FF7 Remake Movie? Not quite. The controversial - and frankly awful looking - DLSS5 has leaked courtesy of an early-access build of NBA 2K27, and modders have since figured out how to inject it into other titles, including the likes of Final Fantasy VII: Rebirth, Control and Cyberpunk 2077 using the RenoDX mod. Of course we have to preface this by saying this is an in-development system and not final, but the core complaints about why it looks like hot garbage remain. The artistic intent is lost, every face is overly lit, reshaped, with dead eyes, and in the cases of most women increasingly sexualised. Meanwhile anything stylised with non-realistic art styles look horrific. Regardless of Jensen Huang’s insistence that gamers don’t understand the technology, I doubt this will have much luck in convincing people otherwise. Speaking of Jensen & co…

With the Ink Still Drying on the Fenris/Deepmind Collaboration, Google have since published a new blog discussing their exploration of games as a lens for their research. It’s a rather truncated version of their history, but worth a read nonetheless!

Nvidia Seeking to Buy Hugging Face? A story that set off alarm bells for me is Nvidia’s intent to drop $12.9 billion to purchase Hugging Face: the open source platform for AI models that grew to prominence during the Deep Learning boom. As I’ve argued before much of the 2010’s surge in AI innovation was achieved thanks to open weights, open models, and sharing of expertise. While Nvidia will be keen to cement their current status as top dog - and have made a lot of noise about supporting open source AI development - it makes me worry as to whether they will continue to support this mentality when their stock eventually begins to drop.

The Man Has Left the Machine: After decades of running the (rather exploitative) market for digital labour, Amazon is shutting down Mechanical Turk - the ‘artificial artificial intelligence’ platform set up back in 2005 to pay workers for conducting digital tasks that AI could not, but often in service of it such as data labelling and processing.

Something Something GTA6 With the big reveal published to Netflix last week and later YouTube, discourse around Grand Theft Auto 6 is in overdrive. Including two headlines that noted there will be no microtransactions or GenAI in the game at launch, and that they had a separate team working solely on NPCs. Though this sounds like it was focussed on aesthetics than actual behaviour. Given both my social media feed, my hectic schedule, and my general lack of interest, I have yet to see this game in motion. I’m happy to wait until November at this point.

“Somebody will make the Half-Life 2 of AI” says Saber Interactive’s Chief Creative Officer Tim Willits. He does admit it’s a silly example, but the headline of his interview with Eurogamer is that a studio will develop a game that showcases the potential of (I presume he means Generative) AI in games. I mean yeah, I’ve said in order for this technology to ever reach a point of maturity in the eyes of players this needs to happen and currently I don’t think it will happen for another few years at least. Meanwhile the collective opinion on (Generative) AI will continue to be poisoned as people use the technology for exploitative purposes. But I also think the Half-Life 2 comparison is rather naïve given that game is the result of numerous technology strands reaching a point of maturity all at once.

Rainbow Six Tactics! There’s no news story here, but as we heard about the games announced at Gamescom Opening Live while enjoying our team dinner the two games that caught the team’s excitement was Stage Tour (which we played, it’s dope), and Rainbow Six Tactics. I love a good tactics game - keen to play Star Wars Zero Company once the dust settles - and I was immediately sold based on this trailer, not to mention that it’s being developed by Ubisoft Paris who co-developed the fantastic Mario + Rabbids games for the Nintendo Switch alongside Ubisoft Milan.

Don’t you find it funny that everyone has a tactics game nowadays except X-COM?

Thanks for reading AI and Games! This post is public so feel free to share it.

Share

Post Gamescom Musings

It’s now been over a week since Gamescom Dev and the AI and Games Conference team helped to host and curate the AI tracks for the event. I am far from recovered from this venture, and after two days of hosting the event while also speaking in other tracks - shout-out once again to Duygu, Matthias, and Gabriel who led the charge in my absence - we still had the rest of the bizdev to conduct from Wednesday to Friday. In short, I’m exhausted, and my workload on returning to my desk has not helped. But hey, sleep when you’re dead etc. etc.

Nonetheless a lot of what transpired during that week is still swirling through my mind, and I figured what better thing to do for this edition than try to collate those thoughts into something more coherent. So here we go…

An Appetite for Meaningful AI Discourse

I think the open question of whether AI will continue to be a point of conversation in games is largely addressed, but perhaps the most useful and meaningful thing I’m seeing is that the conversation has begun to move forward on what all of this stuff really means in context. We know (Generative) AI isn’t going anywhere, but now that it is in the hands of so many people, what does this mean for the landscape of creative industries?

Sure, there is still a lot of (rightly justified) anxieties and concerns - more on that in a moment - but what the reality is we need to learn to work with or around AI. As such, it means people are looking for insight into how AI is intersecting with specific disciplines, both to understand whether this is worth exploring, but also for studios to grasp what the state of the technology is. I can’t help but acknowledge that the vast majority of talks we had in the AI tracks for Gamescom Dev had attendee signups (where people can add it to their intended schedule) that vastly exceeded the capacity of our room. In fact several talks were standing room only, and even my presentation in another track entirely had to turn people away.

There’s a desire to hear from people in their areas of expertise discuss how AI intersects with their field. It’s something we thought about a lot when curating the track for Gamescom Dev as we saw devs from co-development studios, or specialists in areas of tool design and QA pitch talks about their perspective on AI. Not just AI experts talking about their field, but also hearing from subject matter experts about what value if any AI provides for them.

Plus, a quick note to acknowledge how happy I am to see many of my friends and peers from the North American game AI community make the trek to Cologne. It’s great to see that people are willing to travel across and engage with us.

AI Governance Continues to be a Big Issue

I stated but a few weeks ago on my return from Develop that AI governance was a big talking point, and the crowd at Gamescom (Dev) equally shared in those concerns. Many studio leads raised with this with me in conversation, and the need to get on top of these things.

In fact, Shraddha joked that during the segment on governance in my Gamescom Dev talk that the number of people holding aloft their phones to take a photo of the slides made it impossible to see the screen from the back of the room where she was sitting. Nobody has the definitive answer to what AI governance looks like, but it’s clear some studios are much farther ahead than others.

While I’m glad to see people are interested in this - and of course, it’s something we help support in our consultancy activities - for me it was equally rather concerning. Whenever I talk about governance I ask attendees to raise their hand if they either a) don’t have an AI policy in their studio or b) don’t know what it is, and around 80% of the room put their hand up. When you consider that this is at Europe’s leading developer conference - a continent which has significant legislation on AI - I find that rather worrying. We need to get a better grip on these issues.

On-Device AI Will be the Battleground of 2027

Now of course I know we talked about on-device Machine Learning and Generative AI last year given it was a big talking point at the AI and Games Conference in 2026, what’s becoming clearer to me is just that this trend is continuing, but the market is getting busier. More and more companies are investing in this, ranging from the larger players to new (and in some cases ‘refreshed’) start-ups building their own pipelines and toolchains. I had so many meetings with companies who are provided some means to execute on-device AI, be it in pipelines for deployment, to hardware for execution, and toolchains for training. While I think the hardware ecosystem itself has not sufficiently matured at this time, it feels like this is a battleground people are keen to win before most consumers can get their hands on devices ready to utilise it to full effect.

Nobody Has This Figured Out

In amongst my talks, panel presentations, business meetings, and even a roundtable I hosted on behalf of AWS, perhaps the big takeaway is the sense that nobody has this whole ‘AI’ thing figured out. Organisations ranging from indie to AAA are at different levels of literacy, governance, research, experimentation, and deployment. While this may sound concerning, I actually feel rather optimistic about it. Because what I often hear is not just similar same concerns coming from different corners of the industry, but a desire to try and address them. All this wrapped up in a much more mature conversation in terms of AI’s validity and application versus those I was having maybe two years ago.

I can’t get into specifics given we operated the session under Chatham House Rule, but the AWS Roundtable was perhaps the most refreshing conversation I’ve had in the past year or so about AI’s intersection with game development. We had some very senior people from a number of high-profile studios in that room, and had fairly open conversation about not just the concerns and stressors that we’re experiencing now, but the future that we hope will not just materialise, but we will have an active hand in building. The desire to prevent the games industry from devolving into a slop chute was highly evident. People want to explore how to use this in ways that ensure the industry continues to thrive. Nobody has this figured out, but I think we’re a lot farther along in the journey of figuring this out than we were a year or two ago, and in some ways this is very encouraging.

The East/West AI Divide is Becoming Increasingly Apparent

So this is an issue that has been eating at me for a couple of years, and it really was hammered home for me last week. There has been a growing disparity in the level of AI adoption in game development between Europe/North America and the broader Asian-Pacific region, notably China and South Korea but also more recently Japan. Our presentations from speakers from Tencent at the AI track highlighted the level of R&D going into exploring how various AI technologies can be used in both gameplay and production. The attitude, approach, and investment in using AI in game development these Asian countries is quite different from what’s happening in western studios, and it strikes me there is a gap emerging not just in technology, but also in culture around AI in game development.

In the meantime, the level of research and development continues to grow at rates that only a handful of western AAA studios (and research labs) can compete with. My final engagement at Gamescom was to present behind closed doors to a delegation from Krafton as part of their ‘Analysts Connect’ event. I am sadly under NDA as to the nature of the conversations and indeed the data shared by Krafton as part of this, but it further cements my perspective that how we approach AI in game development is becoming another vector through an east/west divide is built.

Now more than ever I feel we need to be working to build bridges, not just to understand the R&D and deployments that are happening on both sides, but also to understand the cultural differences towards AI integration.

Naturally given I spent a lot of time working on knowledge exchange, be it through out content creation, consultancy or the conference, I feel like there’s something to be done here. I dunno, what do you think? An AI and Games Conference spin-off? I hear Singapore is great this time of year!

Wrapping Up

That’s us for this week. I am still trying to catch-up on sleep, as well as get on with work we have planned as I’m back on the road to the US in but a couple of weeks time! More on that soon. But once again I want to say thank you to everyone for their time and energy throughout Gamescom, and a special thanks to our broader team across AI and Games and AI and Games Conference who made this a fantastic week. Time flies when you somehow drink 46 Kölsch in one evening…

See you all next week!

AI and Games is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.