Live data from Hacker News

GitHub Copilot Workspace: Technical Preview

github.blog

121–130 of 333 posts

Re: GitHub Copilot Workspace: Technical Preview

#121
post #84

Interesting. I don't think the AI code generation will live up to developer expectations, but I do see the value in "project management for developers". The value I see with workspace is not the code generation, but the ability to help developers organize their thoughts.

The real problem is a large part of the enormous hype train is investors thinking they can replace talented engineers w/ AI. That's fundamentally doomed for failure. What AI can really do well is take an already competent engineer and suddenly get rid of a lot of the annoying tedium they had to deal with. Whether it's writing boilerplate, doing basic project management, organizing brain dumps/brainstorming, etc.

Pretty much this. I worry that we might face an AI winter soon as investors in 2029 suddenly realise that LLMs are not going to replace anyone anytime soon and that they won't be able to obtain the fantasied cash out of replacing millions of workers with LLMs.

Looking at the sad fate of Deep Mind (R.I.P), I feel that the shortermism generated by LLMs is going to be really painful

Re: GitHub Copilot Workspace: Technical Preview

#122

While I've not used this product, I've created somewhat similar setup using open source LLMs that runs locally. After having used it for about three months, I can say that debugging LLM prompts was far more annoying than debugging code. Ultimately, I ended up abandoning my setup and going in favor of writing code the good old fashioned way. YMMV

Creating this using Open Source LLMs would be like saying you tried A5 Wagyu by going to Burger King, respectfully.

I think benchmarks are severely overselling what open source models are capable of compared to closed source models.

Re: GitHub Copilot Workspace: Technical Preview

#123

While I've not used this product, I've created somewhat similar setup using open source LLMs that runs locally. After having used it for about three months, I can say that debugging LLM prompts was far more annoying than debugging code. Ultimately, I ended up abandoning my setup and going in favor of writing code the good old fashioned way. YMMV

Locally running LLMs in Apr 2024 are no where close to GPT-4 in terms of coding capabilities.

Re: GitHub Copilot Workspace: Technical Preview

#124
Cool idea! Agree on the end vision. The alpha is in figuring out how to get there and making it work really well along the way.

Also, why isn’t anybody connecting the Copilot Workspace announcement to Devin? Biggest company in the world announces a release of a competitor to the most widely seen dev product announcement of the year? Only saw one incidental mention of the connection.

Re: GitHub Copilot Workspace: Technical Preview

#125

While I've not used this product, I've created somewhat similar setup using open source LLMs that runs locally. After having used it for about three months, I can say that debugging LLM prompts was far more annoying than debugging code. Ultimately, I ended up abandoning my setup and going in favor of writing code the good old fashioned way. YMMV

I can definitely echo the challenges of debugging non-trivial LLM apps, and making sure you have the right evals to validate progress. I spent many hours optimizing Copilot Workspace, and there is definitely both an art and a science to it :)

That said, I’m optimistic that tool builders can take on a lot of that responsibility, and create abstractions that allow developer to focus solely on their code, and the problem at hand.

Re: GitHub Copilot Workspace: Technical Preview

#126

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…

Well, I've been out of work for now over a year, and it's the third time in 10 years. People will say that advances in tooling create more work, but ultimately more work is created when there's more money flowing around, and when that money is invested in tooling to eek out productivity gains, which will continue, but will it outpace how much we're padding out the bottom of the funnel? Will so much more money start flowing around for a good reason that it matches how many people can do the work?

It's also worth considering that if you finished school prior to 2020 and started trying to tackle the brutal fight upstream that software development already was, why the hell would it be worth it? For... the passion? For... the interest in technical stuff? Quite frankly, in a tech career, you need to get quite lucky with timing, skill, perception of your own abilities and how they relate to what you're paid to do, and if you have the ability to be passably productive at it, it's at least worth considering other paths. It may end up comfy, or it may end up extremely volatile, where you're employed for a bit and then laid off, and then employed, and laid off, and in-between you end up wondering what you've done for anyone, because the product of your labor is usually at-best ephemeral, or at-worst destructive to both the general population and your mind and body; waking up and going straight over to your computer to crank out digital widgets for 8 hours might seem lovely, but if it's not, it's isolating and sad.

Also worth considering the tax changes in the U.S that have uniquely made it more difficult to amortize the cost of software development, but I don't claim to understand all that yet as a non-US person.

Re: GitHub Copilot Workspace: Technical Preview

#127

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…

Programming languages are a formal way to deliver instructions to a machine. Software is a set of instructions whether you use Python or some visual language . LLMs are just another way of generating those instructions. You still need someone that knows how those instructions work together logically and how to build it using best practices. You can't escape that no matter what you use (Python, some no code tool or LLMs).

So in that sense, the role is not going anyway anytime soon. The only thing that could change is how we make software (but even that is unlikely to change that much anytime soon)

Re: GitHub Copilot Workspace: Technical Preview

#128

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…

The hardest part of software development is not writing code, full stop. It never has been and it never will be. The hard part is designing, understanding, verifying, and repairing complex systems. LLMs do not do this, even a little bit.

I was looking for something like this. The only that might change is some of your toolset but LLMs won't change the nature of the job (which is what people seem to be thinking about)

Re: GitHub Copilot Workspace: Technical Preview

#129

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…

Accountants thought spreadsheets would kill their profession, instead demand for them exploded. 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 mi…

Well said. Assistive technology is great when it helps developers write the “right” code and tremendous destructive to company value otherwise.
Post reply on HN