I'm glad AI didn't exist when I learned to code

Couldn't have said it better. Unfortunately people who should know better like Zuckerberg are perpetuating the idea that coding is the whole job.

> If you can't be bothered to deal with the easiest part of your job, writing some code, how do you expect to be trusted with the hard stuff?

If writing code is the easiest and least impactful thing you do as a software engineer, why does it matter if you use AI to assist with that part? Or to come at it from another angle, why is using Stack Overflow/Google in the hunt for answers and examples good, but using AI models is not good?

There's also the "big picture" part of software engineering, that is the architecture and design of a piece of software. It's all done in the head (and maybe on paper). Once my fingers hit the keyboard the code pretty much writes itself because the hard part has already been done.

True, to a degree but I think this reflects different levels of responsibility and experience as you grow in your career.

Coding is the primary part of the job particularly when you're a graduate, to solve problems using code and you need the skills to be able to code and solve basic problems using code. If you don't know how to code then you will struggle.

As you build that experience and grow in your career you begin to learn and have to capacity to think bigger and start to incorporating aspects of time, money, process and other business factors that you don't really have the ability to do if you haven't spent time in the arena.

I work with a lot of graduates in cyber security and many of them have very "technical thinking" but don't think about business tradeoffs of time and money when looking at security controls. That is primarily because they're still trying to harness their technical skills, which is what they will do early in their career.

Absolutely agree. I went through this as well with friends who did bootcamps during the covid craze. After 3 months they understood how to code and the syntax, thinking that’s all they needed to be a “full stack developer”

When they weren’t getting any jobs I tried to explain that the syntax of coding is the easy part. The hard part is everything else.

Yes, and the persistence to figure things out when they're not working the way you think they should. So many people just go and ask for help the moment things aren't working and if they still can't get an answer they just give up. It doesn't matter if you're coming to me or ChatGPT, if I just tell you what's wrong you never learn.

Maybe I’m missing the nuance here but I don’t understand the arguments against “blindly copying” in the parent post. I agree it is probably not good to give up like the sibling post is saying, but at some stages you kind of do need to copy code blindly, no?

You shouldn’t copy if you’re a student. But if someone is a working Salesforce Administrator trying to learn Apex as a first programming language, I imagine a lot of the introductory code you will see around Apex Triggers will make no sense but you’ll have to blindly copy it if you want to follow along.

Some learners will get frustrated with that teaching approach because they’ll be very curious about the syntax. I imagine at that point they’ll just consult other resources to get their questions answered.

Ironically, as a more experienced programmer, I regret not manually writing every line as I read a book or watched a tutorial, and deliberately do that now. I compare it to thinking I can learn something by listening to a podcast; I can learn about something from a podcast, but to learn it at even a shallow level, I have to have at least more the most shallow interaction with it. Write the code, do it the hard way, then once you've done that, use whatever you want to help you do so more productively. This has coloured my perspective on many things, and has led to me distancing myself from "information" media, because it's often fake learning that costs you the time you could spend actually learning. I don't know where AI sits yet in relation to this, but my sense is that in moments where I'm trying to understand something at a deeper level, it's more fruitful to think about it as hard as possible first for at least a while before looking for the answer. It's not productive in an output per hour sense, but it requires a different intensity of energy that feels like better personal growth.

When I started at Oracle way back in the day, there was no Google.

I was charged with building a DSL so that users could specify their own business logic and tax rules.

I spent 5 x 16 hour days in a flow state coding in C and emerged with a parser that worked remarkably well, but was hideously slow.

After sharing the results with colleagues, one of them told me about lex and yacc. I rewrote it and it went from thousands of lines of code to a couple of configuration files and ran probably 1000x times as fast.

Of course if search engines had been around then I could have skipped all those hard steps and that project would have been completed more quickly.

But the skills and confidence I got from first doing it the hard way really helped in my career.

I don't see it as very different than humans who must follow a narrow script. And scaling support is a good thing. And then there's still prompt injection in lieu of social engineering

Gemini is far and away my least favorite llm. So far I've had OK luck with windows copilot for a simple react site. The site was done in like 2 hours but I polished it up while the person I did it for was copyediting their whitepaper for the "site". I had 5 tabs in notepad++ and was saving it to a cifs share so I could test immediately. React is pretty snazzy but js with its decorations doesn't look quite right to me yet.

I used deepseek-R1-qwen-distill-15B to do a nodejs discord bot. That was done incredibly fast. I had asked for a few days for a cellphone app but llama, openai (didn't test 4o or newer yet), deepseek didn't let me know that what I was asking was technically impossible - service workers without a backend somewhere so I could keep track of timers and fire notifications. Like a PWA.

After asking copilot about it the other way around "can service workers URL be file:///?" Apparently not.

All of the serviceworker nonsense LLM gave me the two-step - make change, revert change, repeat.

These are the people who will be most vulnerable to being replaced by AI.

> ...write code that just kinda breaks the robot and then they’re totally lost for a while.

Um, yeah, that's how I roll.

Try something, it doesn't work, debug for a while and maybe I can get it to work?

My favorite example was I was trying to get the python buffer protocol working for images in blender. I read the docs, wrote some code which didn't work, started debugging for a long while and, finally, after I went digging into python's repo I found out it wasn't even fully implemented. Needless to say I was less than happy they would let a half-baked feature into a release.

I can say the current batch of 'thinking' AIs are pretty good at debugging once you get them pointed in the right direction.