[55yo] My sense is that those problems we worked on in the 80s and 90s were like the perfectly balanced MMORPG. The challenges were tough, but with grit, could be overcome and you felt like you could build something amazing and unique. My voxel moment was passing parameters in my compilers class in college. I sat down to do it and about 12 hours later I got it working, not knowing if I could even do it.
With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
You switch difficulties, like you do in a game. Play on Hard or Survival mode now. Build greater and more amazing things than you ever did before.
We have never, ever, written what the machine executes, even assembly is an abstraction, even in a hex editor. So we all settle for the level of abstraction we like to work at. When we started (those of our age) most of us were assembly (or BASIC) programmers and over time we either increased our level of abstraction or didn't. If you went from assembly -> C -> Java/Python you moved up levels of abstraction. We're not writing in Python or C now, we are writing in natural language and that is compiled to our programming languages. It's just the compiler is still a bit buggy and opinionated!! And yes for some low level coding you still want to check the assembly language, some things need that level of attention.
I learn more in a day coding with AI than I would in a month without it, it's a wonderful two-way exchange, I suggest directions, it teaches me new libraries or techniques that might solve the problem. I lookup those solutions and learn more about my problem space. I feel more like a university student some days than a programmer.
Eventually this will probably be the end of coding and even analytical work. But I think that part is still far off (and possibly longer than we'll still be working for) in the meantime actually this for me is as exciting as the early days of home computing. It won't be fun for ever, the Internet was the coolest thing ever, until it wasn't, but doesn't mean we can't enjoy the summer while it's summer.
>With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
I think it's possible that we'll get to the point where "so can anyone else" becomes true, but it isn't today for most software. There's significant understanding required to ask for the right things and understand whether you're actually getting them.
That said, I think the accomplishment comes more so from the shaping of the idea. Even without the typing of code, I think that's where most of the interesting work lies. It's possible that AI develops "taste" such that it can sufficiently do this work, but I'm skeptical it happens in the near term.
In my experience, the vast majority of people can't even muster the hubris to think they can understand software at such a basic level, it doesn't matter if GenAI did all the heavy lifting sadly.
> With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
That's the thing - prompting is lower-skill work than actually writing code.
Now that actually writing code has less value than prompting, and prompting is lower skill than writing code, in what world do you think that the pay will remain the same?
> Now that actually writing code has less value than prompting, and prompting is lower skill than writing code, in what world do you think that the pay will remain the same?
Don't you think people said the same thing C and Python? Isn't Python a lower skill than C for example?
Great! I turn from a creator to a babysitter of creators. I'm not seeing the win here.
FWIW, I use LLMs extensively, but not to write the code, to rubber-duck. I have yet to have any LLM paired with any coding agent give me something that I would have written myself.
All the code is at best average. None of the smart stuff comes from them.
I think there's still quite a chasm out there. Domain knowledge, an informed and opinionated view on how something should function, and overall tech knowledge are still key. Having those three things continues to greatly differentiate people of equal coding skill, as they always have.
Yeah, but I used to be a wizard with arcane knowledge making computers do things others didn't even understand. I was casting fireballs, and now everyone has Find Greater Familiar right out of the gate who does all the heavy lifting. :(
That’s something LLMs are also presumably good at. At least I’m seeing more and more push to use LLMs at work for ambitious business requirements instead of learning about the problem we’ve been dealing with. Instead of knowing why you are doing what you’re doing, now people are just asking LLMs for specific answers and move on.
Sure some might use it to learn as well, but it’s not necessary and people just yolo the first answer claude gives to them.
Sure but here OP was left wondering why prompting didn't make them feel like they had done/accomplished anything. And the reason is because they didn't do anything worthy of giving them a feeling of accomplishment.
All the ones I've tried feel like little toddlers that completely miss the point, forget half the requirements mid way, are adamant that they are completely correct then have the gall to act an authority when you point out glaring issues.
I take way less time doing it myself vs coaxing an AI to get a decent solution that catches all edge cases.
AI for me is only useful on subjects I know nothing about, and even then, given I know how bad it is in subjects I know everything about I take everything it says with a megacrystal of salt.
I was extending into the future a little bit. Think of it as playing with a cheat code that let's you have more health or power rather than God mode.
But God mode is on the way. ChatGPT mysteriously went from not understanding SAP ByDesign's WSDLs to having fantastic information over the course of a month. The amount of effort being put into AI isn't about the theoretical limitations of LLMs it is how many everyday problems will AI with all the workarounds and hacks ultimately be able to replace mid life career developers?
And that's exactly what the person I was replying to seems to be complaining about.
So many people on "Hacker" News could benefit from reading the canonical text on the subject by Steven Levy. A true hacker wants to bring the fire down the mountain. People around here just want to piss on it.
I suppose I could just tell you what I'm feeling, not the op. The fun was bringing down the fire. That is INCREDIBLY fun! That feels valuable. That feels like you are helping others.
Now, the mountain is gone. All the skills I learned to navigate it are becoming obsolete. Sure, tools get better, new models are adopted, but it didn't wipe out the mountain. When the world move from the 90s to the Internet age, I took my pitons, backpack, and rope and started climbing higher. I felt like I was empowered to bring MORE fire down to MORE people.
With vibe coding getting better and better, I still have my leadership skills, my ability to understand navigation, and all those skills, but the value and joy of climbing isn't quite there. There is a road leading to the top of the mountain and cars go up there all the time.
That is more than just change. That is something the Luddites faced: it wasn't technology and mass production they decried, it was a loss of identity. Kirkpatrick Sale's "Rebels Against The Future" is an awesome history of the luddites and I think it is very relevant today.
With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.