11–20 of 333 posts
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
I just wrote up my experiences of using it for a week: https://austinhenley.com/blog/copilotworkspace.html TL;DR: Copilot Workspace is a great concept. But the UX and framing is entirely wrong and sets the wrong expectation for users. The current prototype is very slow (5+ minutes for one-line code changes) and the generated code is often buggy or has nothing to do with the specification. It doesn’t help me understan…
Thanks for writing that up. It's what I suspected from my use of CoPilot. I love copilot as an autocomplete tool... but it frequently gets things wrong, and using the chat feature to ask it to complete some task usually just generates code that breaks things. So until that improves, Im skeptical a workspace tool would work. Workspace seems like an awesome idea though.. once the tech is further along.
It's a fork of VS Code with some AI features sprinkled in. It writes around 80% of my code, these days.
It also has a few useful features:
- a chat interface where you can @-mention files, folders, and even documentation
- if you edit a line of code, it suggests edits around that line that are useful (e.g. you change a variable name and it will suggest updating the other uses, which you accept just by pressing Tab)
- as you're writing/editing code, it will suggest where your cursor might go next — press Tab and your cursor jumps there
Github copilot trained on GPL-licensed code? Brad Smith you continue to outdo yourself
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
I recently had another round where I tried to put aside my existing thoughts and perhaps biases and tried a trial of Copilot for a couple of days, using it all day doing tasks. Nearly every single piece of code it gave me was broken, and I was using Python. I was trying to use it for a popular Python library whose documentation was a bit terse. It was producing code from the various versions of the library's API, and nothing it gave me compiled. We ended up just going in circles, where it had no idea what to do. I was asking something as simple as "here's a YAML file, write me Python code to read it in" (of course in more detail and simple steps). It couldn't do it. I eventually gave up and just read the documentation and used StackOverflow.
About the only thing I have been able to use it for so far with relatively consistent success is to write boilerplate code. But even then, it feels like I'm using more time than just doing it myself.
And that happens a lot with this stuff. I initially got very excited about Copilot because I thought, shit I was wrong about all this, this is useful. But after that wore off, I saw it for what it is. It's just throwing a bunch of statistically correlated things at me. It doesn't understand anything, and because of that, it gets in the way.
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
Because we've seen similar hype before and we know what impactful change looks like, even if we don't like the impact (See: Kubernetes, React, MongoDB).
> executing large-scale changes in entire repositories in 3 years
Is this actually happening? I haven't seen any evidence of that.
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
Every single time a change like this happens, it turns out that there is in fact that much demand for software.
The distance between where we are now and the punch card days is greater than where we are now and the post-LLM days and yet we have more software developers than ever. This pattern will hold and you would need much stronger evidence than “LLMs seem like an effective productivity multiplier” for me to start to doubt it.
Also don’t forget that 80% of software development isn’t writing code. Someone is still gonna have to convert what the business wants into instructions for the LLM so it can generate Java code so the JVM can generate byte code so the runtime can generate assembly code so the the processor can actually do something.
And lastly, there are a lot of industries that won’t touch LLM’s for security reasons for a long time and even more that are just still writing Java 8 or COBOL and have no intention of trying out fancy new tools any time soon.
So yeah, don’t be too down in the dumps about the future of software development.
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
AI can't prompt itself. The machines will always need human operators.
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
Student here: I legitimately cannot understand how senior developers can dismiss these LLM tools when they've gone from barely stringing together a TODO app to structuring and executing large-scale changes in entire repositories in 3 years. I'm not a singulatarian, but this seems like a brutal S-curve we're heading into. I also have a hard time believing that there is enough software need to make such an extreme prod…
Compilers made it much easier to code compared to writing everything in Assembly. Python made it much easier to code than writing C. Both increased the demand for coders.
Code is a liability, not an asset. The fact that less technical people and people who are not trained engineers can now make useful apps by generating millions of lines of code is also only going to increase the need for professional software engineers.
If you're doing an HTML or even React boot camp, I think you'd be right to be a bit concerned about your future.
If you're studying algorithms and data structures and engineering best practices, I doubt you have anything to worry about.