2 Reframes on AI in Game Dev

So I've been thinking and talking to people about AI and LLMs a ton, and so far I've learned:

AI is the savior and satan. It will bring utopia, dystopia, and everything in between. It's a bubble, it's going to be even bigger than we think, and in 4 years we won't even notice it.

Very helpful info, no doubt.

But in a recent conversation, two new ways of thinking about how AI could impact games were suggested. 

Wanted to share them with you.

If you build games for a living, I don't know what AI's final output will be to our lives. But I think it's wise to at least pay attention as people try to figure that out.

Here are the 2 reframes:

1. From “doing the work” to building a new pipeline

I was thinking of AI as “a faster way to do my tasks," as a tool that integrates into my pipelines doing the boring repetitive stuff. 

It may instead be the thing that REPLACES or CREATES entire pipelines

What happens when you stop looking at the question,

“How do I get AI to do this boring work for me?”
and start asking,
“What system could I design through AI that means I don't have to worry about <X thing> at all any more?”

Rather than AI being another tool to gain efficiency, is it a tool that allows anyone to build helpful tools for themselves or others?

The person I was talking to saw that as one of the biggest wins as they leaned into AI at their studio.

In this world, the dev isn’t replaced.
The important but frustrating work that used to get in the way of their value is now handled almost independently of their workflow.

My simple example of this: could we "train" an LLM to generate marketing material that is solid enough based on our existing assets to not need to peel artists away from the content they are creating in order to support publishing efforts? 

This would still require human oversight, but once trained the LLMs may be able to do all the steps in that process. Humans would be "on-call" or just reviewing things. This is much more than AI just being an efficiency gain that helps the human move faster through their workflow.

This isn't hypothetical either, studios are solving for this problem and avoiding the annoying disruptions that can plague production artists.

It's not foolproof, and done badly could lead to low quality or inaccurate work going live in marketing campaigns, so I'm not telling you to hand the reins over to your LLM, but I like the reframe of AI as a tool that builds tools and pipelines, rather than just a tool that handles bits and pieces of existing pipelines.

2. AI will amplify your studio’s culture (for better or worse)

AI doesn’t magically make a bad studio good.
It amplifies whatever you are today.

If your studio is a ticket factory where...

• People "ship it now, fix it later" all the time

• "More content" is prioritized over "more player impact"

• Success = tickets moved regardless of the game state

Then AI will happily:

• Crank out more code you don’t fully understand

• Mask foundational problems with “speed”

• Dig your tech debt pit deeper and faster

If you've got a toxic culture, ineffective work systems, bad technical oversight, and a poor product vision, AI won't help you. 

And by the way, those are the REAL problems preventing studio success, not an inability to create stuff.

But AI will still have an impact into that broken environment. It will make things worse. It will cause problems, get people even more focused on the wrong things, and create excuses and rationalizations for all the problems that exist.

If you have a healthy culture, a strong product vision, and effective ways of interacting with each other, AI can probably integrate into that in a way that helps aspects of your work move faster:

• Devs using AI to explore options and prototype things faster

• Artists building tools and small agents that unblock others

• Teams spending more time on important decisions and learning

AI as a tool will accelerate whatever you've got going on. If you're already headed for cliffs and ignoring that reality, you might get there faster.

If you're a healthy studio with a solid path and strong focus, it might help you on that journey to success.

Don't view AI as a savior. View it as a catalyst, and be careful what you catalyze.

If you want to go deeper on AI, this week’s Building Better Games episode with Ben Chevalier (Chief AI Officer at Mighty Bear Games) digs into:

• How they use AI without throwing artists under the bus

• What “training models on your own IP” actually looks like

• When AI is a genuine force multiplier vs. tech-debt accelerant

• How leaders can talk about AI without creating fear or toxicity

• Practical first experiments you can run in your own workflow

If you’re building games and want to hear from a studio leaning into AI hard and learning where it does and doesn't help...

Listen to the full episode of Building Better Games here:
https://urlgeni.us/youtube/aiingamedev

Cheers!

Ben C.

P.S. 

AI doesn't discern. Leaders like you do, and it has far ranging impacts on your org. Unfortunately, most leaders are untrained and can accidentally cause a lot of harm by focusing on the right things. If you want to increase the positives you bring to your team and game without breaking things unintentionally, sign up to learn more about the Game Dev Leadership Accelerator here: https://forms.gle/nqRTUvgFrtdYuCbr6