Programming vs AI hype
Posted by milonolan@reddit | learnprogramming | View on Reddit | 12 comments
I want to learn to program without AI, trial and errors, reading documentations and just learn to debug by understanding errors.
It's the part that takes the longest but most rewarding and where you actually learn. But with AI hype and things, and the fact I also started learning programming late, I feel behind, I feel as if I'm not valuable if I don't learn about AI, AI frameworks, AI agents etc etc.
I'm still in my second year of bachelor degree, and have one more year until graduation. But things I've heard, like company doesn't want to hire junior because it's "more expensive" than using senior with AI, I feel like I have to drop the whole "learning" and just start using AI so I can get hired. I recently joined a startup which is an "AI" company, he basically build the entire app with AI, but more advance then I'm using it. Like phases to specify and tell AI where in the code base to look etc and to follow architecture etc. But the code is obviously still spaghetti. I'm however gravitating towards medtech. Is there any hope for us?
ryan_nitric@reddit
The way you want to learn is the right way. Understanding errors, reading docs, and building the mental model of why things work. That's what makes you useful long term. AI is a lot less useful if you can't tell when it's wrong, and it's wrong a lot. "AI frameworks" and "AI agents" aren't things you have to learn to be able to use AI?
milonolan@reddit (OP)
I guess to be able to leverage Ai in other ways?
roger_ducky@reddit
Learn the fundamentals.
Then, read some books on how to delegate as a tech lead.
“Using AI” has to do with learning to delegate.
Essentially, it’s not just asking AI to do things, but to constrain what it could do enough so it can “creatively” solve the problem in the way you wanted without micromanaging everything it did.
Previous paragraph, plus being used to looking at code written by others, is what separates a senior dev from a junior or mid level.
milonolan@reddit (OP)
So I don't need to feel bad about feeling behind in all the AI "skills" I'm not learning
roger_ducky@reddit
There are some syntax things but “proper delegation” is actually way more important. This includes specifying coding standards and a proper development workflow.
Less-Medicine3270@reddit
Learning the things you listed for AI is like a week or two messing around with AI. You're not falling behind by any meaningful measurement.
Learning software engineering is still valuable by itself, and if you're good enough you will find work. However, there is certainly more competition right now and finding a job might be challenging even in a year or two.
If you like this field keep learning without AI, so that the day you do use AI you'll be able to use it better than those who didn't. You'll have to fight for a job, so being competent in the field will give you an edge over others who just rely on AI.
Fuzzy_Job_4109@reddit
you're totally right about being able to use AI better when you actually understand what's happening underneath. i've been fixing cars for few years now and it's same principle - mechanics who learned on older cars without all the computer diagnostics are way better at troubleshooting modern vehicles because they understand the fundamentals
the startup thing you mentioned sounds rough though. working with spaghetti code that was AI-generated must be nightmare for debugging. at least when you write bad code yourself you remember why you did something stupid lol
medtech is actually perfect field for someone who wants to really understand programming deeply. medical devices have strict regulations and you can't just throw AI-generated code at them and hope it works. they need people who can write clean, testable code and actually understand what each line does. plus debugging skills become super critical when you're dealing with something that could affect patient safety
milonolan@reddit (OP)
Yes. Luckily he hired an advisor that had been coding for 15 years so a lot of advice and tickets have been given and created, that's why they also need me now to fix those bugs
kylesureline@reddit
Can you build a web page by hand? Sure, but is it practical? No. Just imagine thousands of HTML devs writing every Wikipedia page by hand. Laughable.
This is why frameworks exist. But it doesn’t render (pun intended) learning HTML obsolete.
I think you can apply the same logic to AI:
Can you build an app without AI? Of course. Should you? Considering bow good it is, probably not. But — and this is the key — AI doesn’t render the skills behind building an app obsolete.
… at least not yet, in my opinion.
No_Jackfruit_4305@reddit
I have 5 years experience (mostly front-end) and over the last year I've been learning back-end (springboot, etc.).
My company isn't giving us a choice about using AI, and it is getting in my way. Sure I get the work done, but I am writing very little code. My learning process has slowed. If it were up to me, I'd be pair programming with a senior and they would tell me which docs/manuals to read.
So don't worry about AI OP. Lean on your more experienced friends, maybe go to office hours.
ZestyHelp@reddit
"learning AI" is something my grandmother was able to do in an afternoon and she can't even delete apps on her phone so ..
NervousExplanation34@reddit
learning AI is easy, learning programming is hard.. AI is also still evolving while the fundamentals of programming don't. Why the fuck would learning AI first be the better choice? Isn't the value of a human to do something the AI can't, rather than to know how to use AI?