Afraid
Posted by Recent_Land1282@reddit | learnprogramming | View on Reddit | 47 comments
I am new to software engineering and see the immense speed with which ai is growing. I cannot afford claude code or codex or stuff that is paid. I also want to be a good software developer and land a good job. Everywhere I see people using ai and making everything so quickly, i don't know how to do this, can anyone help me to also catch up with the updates and be as productive as everyone else here?
I also want to contribute to open source but the easiest first issues are taken by people who use ai. Am I doing something wrong or shall I just quit?
pepiks@reddit
AI is hype. It is good for basics, but when you learn basic you will not need it. Anyone who think that agentic coding replace developers is mistake. Only real shift is - you have be more educated than previous generation and be more on level old schoolers from pre80s and 90s to be more sure that you will not lost your job.
IT is hard. Programming is fun and hard. It is easier way to make living, but at the same time programming is fun. First make code a new toy. After some time think - I feel it or I will be used it in other context (programming as support other job)?
Recent_Land1282@reddit (OP)
You are correct about that but how can I make it a toy under all this stress?
pepiks@reddit
Make hobby outside programming and... live. Programming has more tense, because AI change thinking about how valualable programmer is, but at the future people with solid basic will replace start current seniors. If someone only has Clause / Chat-GTP knowledge it will stuck on some level and be worse. You have to be ready learning new stuff until you retired and at the same time improve social skills to not be forced to use something fancy-name-shining-new technology.
How make toy is hard and easy at the same time. You have to think not about impostor syndrome, but each day learn new stuff and move on. You can not learn all knowledge on the world. Sleep, eat healthy, add good habbits to maintance work-leasure balance, avoid distraction, make things meaningful and it will be OK.
Recent_Land1282@reddit (OP)
All you said is true, maybe i should just get started and close this tab
aqua_regis@reddit
What good is a power tool in the hands of someone who can't even use basic ones?
This is what AI is - a power tool. In order to use it, you need to have solid fundamentals and a good grasp of programming.
So, you need to learn programming before becoming able to properly use AI to your full advantage.
Contributing to Open Source is not a "must", it's and always should be a "want" and there you should only contribute to applications that you have become attached to through usage and where you can see improvement potential. You don't just pick a random project that you don't have any use for just to list 'contributing to open source' on your cv.
Recent_Land1282@reddit (OP)
But the power tool you are talking about is getting 10x better every month. Me getting a job lies 4 years from now.
cylonrobot@reddit
>getting 10x better every month.
I recently decided to learn python, with the help of Claude.
I had Claude create a function for me. The function needed to create an instance of a class for me. Claude's function created two instances but only used one. It was written in a way that a beginner programmer would have.
Recent_Land1282@reddit (OP)
Hmm, that might be true but I use the cline extension in vs code with the free gemini api it doesn't make such mistakes , infact i was fascinated by how good it was for being free. Is claude not as good as that?
aqua_regis@reddit
Yeah, as a beginner, early learner who only does projects of very limited size and complexity it can be impressive. Yet, if you scale up in size and complexity, AI is still very much at its infancy.
Also, the LeetCode like stuff can easily be solved by AI since it has basically sampled the answers to all the questions around.
Come something completely new and unique it will have problems.
You are gravely overestimating what AI is capable of.
Recent_Land1282@reddit (OP)
So should I go forword with being a swe?
aqua_regis@reddit
There are no reasons not to.
If you become good, you will find a job. If you only learn just enough to pass, you will struggle to find a job. Yet, it has always been like that.
Recent_Land1282@reddit (OP)
You are correct in saying that
aqua_regis@reddit
And so? It will still need a skilled operator.
No, it's not getting 10x better every month. That's only what the people trying to earn money (yes, all the AI companies are in the red figures and some are even close to going bankrupt) want to make you believe.
There are certain "quantum leaps" if you want to call it that, but even that goes only so far.
Recent_Land1282@reddit (OP)
True but why are companies like cognizant laying off currently then?
aqua_regis@reddit
Because of the hype. It has always been like that in one way or another.
Something new pops up, gets hyped, affects everything and once the hype blows over, people who actually can work are hired for double or triple the rates to fix the mess.
Recent_Land1282@reddit (OP)
But people say this is more of a shift than hype. You are saying they will hire at higher rates but it seems that will be not the case.
aqua_regis@reddit
Yes, at the very end it will become a shift. Right now, it's a hype.
Management and stakeholders think they can replace programmers with AI for now. They will sooner or later awaken when they find out that there will be no one left to maintain and fix the produced garbage.
Similar things have been going on for long long times.
Web Dev had been pronounced dead when the first drag&drop web builders (e.g. Dreamweaver) came up. Then, again, when the first CMS (WordPress, Joomla, Drupal, etc.) came up. And so on, and so on. Yet, there are still jobs.
Similar in accounting. It had been pronounced dead with the advent of calculators, then with the advent of spreadsheet programs. Still here.
Don't believe everything the news tell you, especially not when information comes from people earning money with it.
Yes, there are layoffs, but that started already post COVID and pre-AI. AI only has increased them for now and now they are more public.
AI is not the magic "solve it all" and it will never be.
If AI grows like it does now, predictions say that there won't be enough energy globally to supply it. Hardware is limited, and currently close to unavailable and absolutely unaffordable because the AI centers buy everything.
The whole is bound to plateau or even crash. It is not sustainable in the long run.
Even if it continues like that, the costs to keep AI running (that already are astronomical) will explode, which, in consequence will be pushed to the users of AI making AI less and less affordable. Right now, the companies are pushing AI to the broad masses to create dependency and a market and then, they will increase the prices (they absolutely need to) in order to at one point become profitable.
InfectedShadow@reddit
Actually it's more like 2x every 7 months according to recent studies.
Recent_Land1282@reddit (OP)
Still pretty fast
SharkSymphony@reddit
> getting 10x better every month
No, it's not – not even close – though I understand that, from the outside, it sometimes looks like it. Go look at the Claude sub – they're freaking out about whether the new model is even 1x as good as the prior one!
FOMO in tech is nothing new. The advantage to mastering the fundamentals first and not worrying so much about AI now is that you get to skip some of the twists and wrong turns and product instability and broken tools and almost complete lack of guardrails and _nonsense_ the rest of us are dealing with. If we get even _one_ 10x improvement in the next 4 years, why then you can slot right in and take advantage of it. 😁
Recent_Land1282@reddit (OP)
Thanks for the response, is this really true? I also saw people disappointed about the new model by claude.
Astronaut6735@reddit
It's basically advanced autocomplete that gives the statistically-best answer to a prompt. For easy and average things it's pretty good. But it doesn't have discernment, taste, judgement, etc. You still need to decide what is actually going to be the best way to solve your customer's problems, specify what you want it to create so unambiguously that the LLM can create something useful, evaluate what it creates, decide if the implementation is going to be flexible enough to support future requirements that you know about but it doesn't, test/verify all the special usage patterns and scenarios are handled, make sure error modes are handled correctly, etc. AI needs a skilled person to use it.
Claude Code is $20/month. There's nothing you can forego to afford that, even for a month or two to give you time to learn how to use it? This is your future we're talking about.
Recent_Land1282@reddit (OP)
Brother I am broke , my wallet has flies as you see in cartoons.
heisthedarchness@reddit
Wait six months for the money to dry up. LLMs are a fad that you're better off ignoring in order to improve your own skills.
Recent_Land1282@reddit (OP)
You are true for that they just might burn too much, but their contribution is not going anywhere, it has already disrupted the market. They might be replaced by local LLMs or companies like Google who have a budget so immense it just doesn't seem to run out.
Key-Kaleidoscope9859@reddit
Geht mir genauso. Umschulung zum Anwendungsentwickler beendet 2024 und seitdem Arbeitslos…
Recent_Land1282@reddit (OP)
But you graduated in a time when the market was doomed , literally lowest of all time but now it is emerging again, yes there are layoffs but hiring is also going up it's like they are changing how they work. I don't wanna regret later if I pivot today that I might have been a good swe as per the future requirements. Cause my journey will be long from here like 4 years study,2-3 years exp then finally get in the market when it is stabilized. It might look very different from today and I might just live good? As the total time is like 7-8 years.
money_enthusiast123@reddit
In my opinion, what’s happening now with the tech field is akin to whats happening with doctorate degrees in pretty much most fields. Of course pursuing a doctorate has always been for people truly passionate about the subject, but its especially true and relevant now because there’s so much saturation and competition, and the opportunities are much scarcer, so if you want to be successful you need to ask yourself these two questions. 1. Do i truly like the subject and not just want do it for things like prestige and 2. Am I willing to put in extra effort to make myself stand out in the field which takes both hard effort and time. If the answer to both is no, then maybe look for a different field.
Recent_Land1282@reddit (OP)
For both your questions my answer will be yes , I never wanted too much money or stuff just a simple job and I am willing to put in the extra effort. I started learning web dev(basic html,css and js) when I was 14-15 years old I really like it and the new things I hear about like system design and all that, but I fear for my efforts going in vain learning all this and then just boom new model comes up that does everything.
Made-In-Slovakia@reddit
First thing is to learn software development. And for that you do not need AI and even I would suggest not use it during this period for development. Learn language, how it works and why things are done one way or another.
Recent_Land1282@reddit (OP)
Thanks for the response but do you think it will really help as ai is getting better and better and maybe in 3-5 years it will be good enough to outperform all my efforts.
Made-In-Slovakia@reddit
So I am afraid IT is not for you then.
Recent_Land1282@reddit (OP)
So should I stop overthinking and work? Or as you said just leave? I like software development in general this is the only field I have the most interest in and have not thought of anything else.
Made-In-Slovakia@reddit
Based on your other comments, at this moment yes, you are overthinking it. First, this "AI" as now is not magic-everything-will-do like in movies. It is tools that needs to be used by skilled person (I call it "sh.t in, sh.t out"). Second, if you really like it you should do it. Do not think about uncertainty of future, especially one that you can not predict or affect.
If you like software development then do it. Now it is hard time for entry levels but you have to be persistent if you are really interested. Became software engineer, someone who actually understand systems, security, architecture, etc. SE is not only writing code, for senior roles it is really may be 25% of their work. These AI code generators will replace all-day code monkeys that calculates their performance on number of lines of code, but not true engineers.
And one recommendation, stop reading crap from people/companies that profit from this AI FOMO. For each article about greatness of AI I can give you one about how people/companies lost money because AI slop.
Recent_Land1282@reddit (OP)
You are correct in what you are saying and i should work as I can't control the future. But everytime I work this fear comes again and breaks my consistency. I think that's for me to figure out.
canIkick1it@reddit
This might be an unpopular opinion but yes you should probably find a different career. Computer science grads are having a tough time getting entry level positions, the work is easily out sourced and like you said AI is making it incredibly accessible
Recent_Land1282@reddit (OP)
Thanks for the response, and as you said about the outsourcing I am not afraid of it as I live in a country where most of the outsourcing happens so that is not a concern, the problem is that even getting a job in these outsourcing companies will become very hard if the improvements in ai continue like this for 2-3 more years. However I sadly just cannot leave it's like a toxic relationship even if I do anything else business,finance or something I get drawn to software development every other day but when I am thinking about it as a career I am afraid, it's like a paradox.
canIkick1it@reddit
If you’re passionate and can work and finish compelling personal projects then go for it I guess.
Recent_Land1282@reddit (OP)
Man this is the problem I am not getting started due to all this overthinking
Dazzling_Music_2411@reddit
Relax, you've only had a Reddit account for 6 days. If you're panicking so much before you've even started, maybe it's better for you to give up and find something less stressful. I can guarantee you it will get a hundred times more stressful in the future.
Recent_Land1282@reddit (OP)
Thanks for the response, you are right to say that I am panicking so much but all I am thinking about is the final result I will get for my efforts, every field is hard and nothing is easy, however one can atleast think of rewards that they will get. I thought about other things too but it seems programming is the only one which is easiest for me as I atleast have some knowledge of it.
Dazzling_Music_2411@reddit
Why does the thought of rewards make you panic?
Recent_Land1282@reddit (OP)
Cause I think if ai is going to everything I might not even land a decent job which I am hoping for that is all the reward I want for my hard work.
Dazzling_Music_2411@reddit
That's very possible.
It is certainly happening to a lot of people, even today, never mind 4 years from now.
If you have doubts in your abilities, it's probably best to start looking for something else from now. There will be programming jobs in the future, but only for high-level experts. It will be very hard, unless you enjoy (and are capable at) high-level challenges. The days of the "code-monkey" seem to be over. In future code might be totally generated by machines, with humans only supplying the specification. But writing a correct spec is advanced graduate stuff, not for everybody.
Recent_Land1282@reddit (OP)
I like all this stuff like system design(mostly lld),dsa and ML, these people say is going to be valuable and stuff, coding was already mostly out of question since the launch of ai.
bootyhole_licker69@reddit
don’t quit man, you’re just in that early panic phase we all hit. learn basics really well first, ai or not. you can use free tools like copilot chat trials, ollama locally, chatgpt free whatever. dodge “good first issue” race and fix docs, tests, tiny bugs instead, that stuff still gets merged. real value is knowing how to read code and debug, ai just speeds up typing. problem is you’re trying to learn in a time where finding a dev job is way harder than before
Recent_Land1282@reddit (OP)
I fear because I see layoffs everywhere cognizant just laid off more. Will working this hard learning all the basics and using ai payoff in next 4 years when it's my turn to land a job. The difference between the capability of ai 2 years before and now is so much more. Just think about 4 years from now.