You say, “Recent analysis suggests that by 2030, artificial intelligence will be capable of creating AAA video games with nothing more than a prompt and final sign-off.”
I remember that someone had instrumented Claude to use Blender and create models (probably through MCP).
I wonder how close agentic systems based on present day LLMs can get to creating a AAA games, if they can either create models, or buy high quality 3D models, music, etc.
Or, if you gave them what a human would likely get... A basic template project to build upon.
The main headache would be enabling testing. So, it'd need to instrument the game in a way that it can more easily interact with it.
... Sounds like a potentially interesting problem to try and solve.
The AI Game contest recently that required 80% AI Coding, was fundamentally decent. Just needed good models.
The real question might be: "How far could you get with a game building system, if you weren't constrained by compute costs, and put 6 months into it?"
You say, “Recent analysis suggests that by 2030, artificial intelligence will be capable of creating AAA video games with nothing more than a prompt and final sign-off.”
What’s the analysis source? Just curious …
I remember that someone had instrumented Claude to use Blender and create models (probably through MCP).
I wonder how close agentic systems based on present day LLMs can get to creating a AAA games, if they can either create models, or buy high quality 3D models, music, etc.
Or, if you gave them what a human would likely get... A basic template project to build upon.
The main headache would be enabling testing. So, it'd need to instrument the game in a way that it can more easily interact with it.
... Sounds like a potentially interesting problem to try and solve.
The AI Game contest recently that required 80% AI Coding, was fundamentally decent. Just needed good models.
The real question might be: "How far could you get with a game building system, if you weren't constrained by compute costs, and put 6 months into it?"