11–20 of 316 posts
{ "error": { "message": "The model: `gpt-4` does not exist", "type":
"invalid_request_error", "param": null, "code": "model_not_found" } }What the articles like this miss is that we understand pretty well how ChatGPT writes code. But what about maintenance, and particularly debugging? How would it make these tasks. And we know that these things take much much more time than writing code.
It's almost useless for high-level design and architecture. Those tasks do take more time than writing code, but they're super-charged by rapid prototyping, which ChatGPT excels at. For example, right now, if I'm picking between three libraries, I can have working versions of my code implemented in all three in a few minutes, and built out a bigger mockup in a few hours. I can make a more educated decision from there. In the olden days, that'd be a few hours and many days, respectively.
It also does really nicely for code reviews, which results in better code. That also speeds up maintenance and, later, debugging. That's especially true in domains I'm less familiar with.
Some maintenance, it excels at. I'm the author of a major piece of open source infrastructure, which is widely used, well-architected, but every technology used is obsolete (JQuery, older Python web frameworks, legacy database, etc.). I feel like I could rewrite the whole thing in React+Node+etc. in a few weeks. There is close to zero architecture work -- it's mostly having GPT write -- or sometimes just translate -- code piece-by-piece.
I don't feel like GPT-4 obsoletes software engineers as it's used yet, but it does really raise the bar on what's possible to build.
does not work for me: { "error": { "message": "The model: `gpt-4` does not exist", "type": "invalid_request_error", "param": null, "code": "model_not_found" } }
does not work for me: { "error": { "message": "The model: `gpt-4` does not exist", "type": "invalid_request_error", "param": null, "code": "model_not_found" } }
I tested the scripts and they work really well, esp like the image generation script. However unlike the web-browser the scripts does not keep the context, which is not great.
is there a way for shell script to maintain context session somehow?
I was working on a .pkg MacOS installer and could not figure out how to get my background image @2x.png to render on a retina display. Turns out I needed to create a directory with the .imageset extension. My attempts at searching for a solution online never led me to this, but ChatGPT immediately knew what to do when I described my problem.
It's also tremendously useful as an interactive manpage. I've found it's much faster to ask ChatGPT how a complicated CLI tool works than trying to read the dense --help output to find what I need.
It can even have a conversation about A* pathfinding, and how best to implement it with different nuances. It's crazy.
I wonder how employers will handle this trend, which is happening so quickly. Is it OK for an employee to get advice on their work from ChatGPT? Is it OK to commit AI-generated code? Is it OK to show internal source code to an AI to help them find a bug or refactor it? There are security implications here, and ethical questions. I think most sophisticated companies will only adopt this kind of tech when they can self-host it, or at least have some kind of guarantee that their employees are using a dedicated, isolated model.
With that said, I'm sure many people are already making heavy use of this for their work (as are students). It's already an out-of-control phenomenon.
Being efficient at writing code can barely make someone a 5x engineer. A team of people who write code effectively is easy to do. Hyperproductivity in software is all about deciding what problems to tackle. Richard Hipp isn't worried about chipotle restaurant orders in golang, he's worried about how to store data reliably. That isn't a coding puzzle that ChatGTP is likely to help with. Either ChatGTP can do the whole…
In relative terms, the unaided 10x engineer is now only a 2x engineer as his peers are now vastly more productive.
That's huge.
Being efficient at writing code can barely make someone a 5x engineer. A team of people who write code effectively is easy to do. Hyperproductivity in software is all about deciding what problems to tackle. Richard Hipp isn't worried about chipotle restaurant orders in golang, he's worried about how to store data reliably. That isn't a coding puzzle that ChatGTP is likely to help with. Either ChatGTP can do the whole…
Still a good post and reply.
10x engineering comes from someone who is trying to "be a better developer" actively for many years. Most people who just use the tools to create the product won't ever become that much better. They will just become more knowledgeable. It is the same as any natural skill. You can play piano for years and never really get better. Or you can try specifically to do hard work of improving your technique, theory, and performance ability directly. It will be way more emotionally taxing to do so, but the end result after years of doing this is that you're vastly better than most. Not everyone can even do this. You definitely have to be in a growth mindset and not particularly stressed emotionally at the outset and during the process.
My coworker at Amazon was just making macros nonstop for all these things he wanted to automate within his workflow. He would focus specifically on the hardest problems that we haven't solved first and everything else around it would just spring up together rapidly (I don't even know how he did this). He ended up being heavily involved with AWS and later went to Netflix.