The trick is to split the code into smaller parts.
This is how I code using ChatGPT:
- Have it analyze how to structure the program and then give me the code for the outline with not yet implemented methods and functions.
- Have it implement the methods and functions one by one with tests for each one.
- I copy the code and test for each method and function before moving on to the next one So that I always have working code.
- Despair because my code is working and I have no idea how it works and I have become a machine that just copies code without an original thought of my own.
This works pretty well for me as long as I don’t work with obscure frameworks or in large codebases.
So my job (electrical engineering) has been pretty stagnant recently (just launched a product, no V2 on the horizon yet), so I’ve taken my free time to brush up on my skills.
I asked my friend (an EE at Apple) what are some skills that I should acquire to stay relevant. He suggested three things: FPGAs, machine learning, and cloud computing. So far, I’ve made some inroads on FPGAs.
But I keep hearing about people unironically using chatGPT in professional/productive environments. In your opinion, is it a fun tool for the lazy, or a tool that will be necessary in the future? Will employers in the future be expecting fluency with it?
Right now it’s a good but limited tool if you know how to use it. But it can’t really do anything a professional in a given field can’t do already. Alhough it may be a bit quicker at certain task there is always a risk of errors sneaking in that can become a headache later.
So right now I don’t think it’s a necessary tool. In the future I think it will become necessary, but at that point I don’t think it will require much skill to use anymore as it will be much better at both understanding and actually accomplishing what you want. Right now the skill in using GPT4 is mostly in being able to work around it’s limitations.
Speculation time!
I don’t think the point where it will be both necessary and easy to use will be far of tbh. I’m not talking about AGI or anything close to that, but I think all that is necessary for it to reach that point is a version of GPT4 that is consistent over long code generation, is able to better plan out it’s work and then follow that plan for a long time.
I just use it for snippets - “here’s my function, how would I go about changing x?” Or, “here’s my block of code, I’m getting this error, what am I missing?” (I know, I’m fine to share my code but not company code)
You can host a model locally with gpt4all. So using company code shouldn’t be a problem, since it wouldn’t leave your machine.
