Hacker News Reader

Comment by softwarebeware | Hacker News Reader

I'm out almost immediately. The first comment is more text than the code that it produces.

That happens sometimes when you move up an abstraction level. I bet "self.count += 1" is a lot longer than the machine code it generates.

Falls apart when we get down to binary, but holds across most levels of abstraction

Only if binary is represented as ASCII, which is an unfair comparison

Think of GitHub copilot as StackOverflow on steroids -- a quick way to write code when you're not sure how to achieve what you're trying to do.

After all, "How to parse a CSV file in Python" is longer than "csv.reader(file)" but without knowing that "csv.reader" exists, you have no other way but to tell Google what you need.

> Think of GitHub copilot as StackOverflow on steroids

This is how I already think of co-pilot, but these steroids seem to be mostly for prototyping.

SO often have comments and context such as "this works with 98% of browsers", "this isn't recommended, try X instead", "this works but can break library code because it changes the global scope", "this stopped working in version X" etc etc. Context like this can be important to take into account depending on what you're building.

Start a line with

    // CAVEATS:
    // POTENTIAL ISSUES:
    // Above is deprecated. Use below code instead

and ask Copilot to auto-complete.

// this should never happen

    fn classify-toxicity(text: string) {
      do_work(text)
      while true {}
      // this should never happen
    }

Soon we can all quit our programming jobs and become managers managing copilot by writing copious amount of comments to persuade copilot into generating our software products. Finally no coding required!

I think you're missing the point: this is starting to open the door to people who can't code.

I think the point is if you cannot/do not know how to code you cannot confirm what co-pilot is doing. Especially when it comes to complex topics like drawing context from natural online language using machine learning.

That's where testing comes in, no?

Entering the programming field then becomes an iterative loop of you instructing an AI to generate code, generate tests and iterate / re-adjust until it does what you want.

"You" as in the next generations of programmers in a decade or two.

I adore copilot and use it daily, but I'm pretty sure if I had always depended on it, I wouldn't be able to properly parse correct from incorrect programs.

It's a really really cool tool and a lot of these comments are just shallow dismissals from people who haven't actually used it and like to be reactionary on the internet because that's the world we live in apparently. But I think it works best when it's used by people with experience.

Hopefully future models with higher accuracy and research in grounding can get us to that point however.

Reply on news.ycombinator.com