Comment by Voltage

6 days ago

I'm on the side of AI is a huge productivity booster (by my guess... 10x)

But I don't want to make the claim lightly, so I did an experiment.

I signed up for copilot pro, and have been using their 'edit' feature. This is more than just their auto complete.

I set myself a goal to create a playable web game of classic frogger.

It took 4 hours with copilot "edit" and my full attention.

I didn't write a single line of code, but I did ask it to refactor and gave it a project description.

I suspect this would have taken me 4 days full time to get to this level.

Try it out:

https://dev.zebrar.com/jd/frogger/

5 comments

Voltage

Reply
guax  5 days ago

So less than 10x already.

Question, how familiar are you with the used technologies? My experience on where Ai have been useful so far is things I don't have a good understanding on but, when I do, its a different ball game, mostly because coding it directly seems faster since I know exactly the behaviour I am looking for and I am not having to deal with unintended consequences.

I see it as the Alice cat thing, when you don't know where you going, any road will take you there. So its been great for exploratory work and prototyping.

  • Voltage  5 days ago

    Yeah, I'm very familiar with the tech, I've been interested in games dev and web dev for a few decades now. So you could be right, that the models aren't ready to "play on their own" yet.

    I tried doing a warcraft 1 clone, but that felt too complex for the model being used (openai 4.1). That model was just the default setting in copilot.

    I dug a little deeper this morning, and it turns out I hadn't actually enabled my copilot 'pro' mode, which has granted access to some more current or dev focused models. So I'll take them for a spin to see what they're capable of.

    My goal here is to roughly get a sense for when a task is too complex for an "agent" to handle.

    I also want to try adding custom tools to suit certain project needs. For example, Unreal Engine has a python editor interface, so I'd like to have an agent drive the editor to build something. I have my doubts.

    Once I have a feeling for what level of complexity can be handled, I'll see if I can manage the tools better using this understanding, by breaking large and complex projects into appropriate chunks of work / complexity.

    /endBrainDump

GenshoTikamura  5 days ago

> a playable web game of classic frogger

You could create a Tetris in even less time. Try to create something more novel and less hot on the heatmap of paths walked in gamedev.

  • Voltage  5 days ago

    I will. I'm just trying to evaluate the tools on a medium size task. I'm trying copilot's agent mode with the goal of creating a warcraft 1 clone. Or something resembling it.