Robert Tamayo


Forget About AI for a Second

Forget about AI for a second, and imagine a tool that can generate code for anything you want. Let's say it's powerful enough to do the wildest things, like generating a video game on a prompt. "Make a 2D Sonic game with a Squirrel instead of Sonic", "Make a 3rd person shooter set in a post apocalyptic world overrun by zombified old household appliances", etc. 

Now imagine that you have to get a little more specific about what you want. Let's say the first output of the tool is not exactly what you had in mind, so you iterate over it a few times, refining certain details.

Let's say you keep going like this until you're describing smaller and smaller details, such as the inventory management system and updates to experience points gained from enemies and so on.

Let's say by the end, you've basically provided so many details that you've essentially written a small novel, but the only thing you didn't do was write any code.

That's practically how things work today, just with programmers, level designers, artists, and others. And that's the stated goal of AI tools - to provide tools for generating code based on prompts.

The prompts aren't much different than design docs, project requirements, or even tickets.

Now that we've forgotten about AI being the implementation of such a tool, let's also forget about the programmers themselves being the implementation of the tool.

In fact, let's forget about the tool altogether.

The reality is that the ideas are important. The game must be fun. The product must be useful.

I imagine that AI researchers ultimately realize this, and they probably think that genius and idea generation can be done by a computer billions of times faster than humans can do it. That's the dream, at least.

Leave a Comment