Andrew Chen Archives

Grand Theft Auto 6 and the future of AI
If you happened to miss it, a few weeks back, here is the game trailer for Grand Theft Auto 6. It’s worth watching, and is amazing on multiple levels. But GTA 6 might be the peak of the open world category, untouched by the next wave of tech, particularly generative AI.

Let’s start some facts: First, GTA has a reported budget of $1-2B, making it THE MOST EXPENSIVE GAME EVER

Not only is this the most expensive game ever, but compare it to movies. The most expensive movies, modern installments of the Star Wars and Avengers and Pirates franchises, clock in at a mere $300-450M. So this is nearly 5X:

I also think this is indicative of the pole position that gaming is taking in culture.

The recent trailer is now the most viewed non-music video on YouTube following 24 hours after its release, beating any movie trailer, TV show premier, and it topped MrBeast. In the first 10 hours, it hit 70M views, and at the time of this writing, a few weeks later, it’s at over 140M

It’s also funny to compare this to building a product in tech. Rather than “move fast and break things” instead this game:

• took $1-2B to build, as we said
• started dev in 2014, so it’ll be 11 years from start to release
• 1000s of developers, designers, etc crunching to finish

But as most of us in tech have been following, the tools and approach to games is rapidly changing.

How generative AI is changing the games industry
We are seeing generative AI hit multiple parts of game development. It’s early days, but there are quite a few places where this is hitting — this includes everything from concept art to assets within the game, to interactions with NPCs:

• creating infinite varieties of concept art
• designing/creating 3D assets
• LLM powered NPCs
• generating environments and worlds
• synthesized speech for in-game characters
• bots to play against, to onboard into PvP competitive games
• endless quests, narrative stories, etc
• etc

But that’s just the “weak form” innovations that are easily imagined today. The “weak form,” as my colleague Chris Dixon talks about, often comes alongside a strong form version, in a pair. The weak form is more easily understandable by the market, but it’s often the strong form that ultimately makes the bigger impact.

The “strong form” AI innovations will impact GTA in more emergent formats. To take a metaphor, it’s been cool to see modding allow for the emergent GTA RP (role play) community emerge, allowing for new game play as people play as cops, gang leaders, and other folks. Millions of people have tried this format of GTA, and even more millions have watched. It’s a new inventive form of play that didn’t previously exist.

I think the same thing will happen for future editions of the Open World genre. Yes, generative AI will be used to make games like GTA more cheaply ($1-2B is a lot!) or to get more content with the same dollars. But also AI will unlock new forms of gameplay as well.

I’m so excited, for example, about what happens with the GTA version of AI town — where NPCs have their own inner voices, motivations, and needs. You could imagine this underlying platform being able to power next gen social, dating, or otherwise. You could imagine thinking of open world games like GTA almost more like a physics engine, with a layer of modding and AI built in. And at some level, it’s a large enough playground that you can build a lot of other game genres inside of it (as Roblox does).

It’s just as likely gen AI will reinvent genres, not just make it cheaper to build
Think about what happened in the last content revolution, where user-generated platforms like YouTube and TikTok allowed video creators to dramatically reduce the cost of content. What creators used the technology for wasn’t to try and compete with Hollywood. People aren’t disrupting the 2 hour film or the 10 episode TV season with short videos. Instead, you see completely new video formats that are native to the medium, whether it’s personality-driven vlogging, video game streaming, long-form podcasts, or otherwise. These don’t compete with Hollywood — they take on entertainment by a wholly other approach, stitching together hours of entertainment 6 seconds at a time.

All this time we’ve been talking about big open world games, I actually think the best new experiences won’t resemble Grand Theft Auto at all. Instead, we’ll see new game genres that are rapidly released that compete in different ways:

• Perhaps we’ll see games as a new format of meme. If a funny presidential debate moment happens between two candidates, perhaps later that evening you’ll see a fully-fledged fighting game (built in a no code, AI-enabled environment) become the huge hit later that evening
• Perhaps gaming will be ultra-personalized, and people will make huge, immersive, deep games for their 50 person college club — just because they can
• Perhaps gaming will be sold alongside commerce and other experiences. Today, you might build a huge gaming experience to accompany Harry Potter, but the economics don’t work to make it to promote your tiny Shopify store. If content creation costs fall close enough to zero, you might.

Either way, it’s really hard to imagine a big budget game like GTA happening the same way it is happening now. Instead, the content will be AI generated, along with the quest and maybe even the genre. And maybe it won’t look like what we consider a game today.

This last year has been a period of exploration for the games industry, but it’s still very early. Everyone is experimenting, which is a good start. And some tools, like Midjourney for concepting, has taken hold. But very little is actually in production. There’s a big transformation coming, and it’s going to be as big of a wave as anything we’ve seen in this industry.

PS. Get new updates/analysis on tech and startups

I write a high-quality, weekly newsletter covering what's happening in Silicon Valley, focused on startups, marketing, and mobile.

Leave this field empty if you're human: