Ask HN: Learning programming when AI can do better?
I'm a beginner at programming and following tutorials and online courses, but now all my friends and especially my big brother all told me to stop learning programming because AI can do any software engineer job and it'll keep being better and better. Is this true? Should I pursue some other path if the thing I like is gonna be automated by AI anyway?
18 comments
I think it's a very interesting point / question. Programmers are wondering, artists surely are too. I think it's sane to have the question in mind.
The capacity of technology is very hard to predict (the future is).
Disappearance of programmers. Right now, you have a lot of programmers, they are not out of job. At the current capacities, they will not be out of job ; to compare with art, I think that pure illustration is a done game. Most prominent people stating that programmers will disappear have vested interests in the AI hype.
Replacement vs enhancement. Right now any programmer is probably using AI. The thing it has impacted is problem solving and information retrieval. See https://www.reddit.com/r/programming/comments/1592s82/the_fa...
Side knowledge. Programming (and sometimes solving technical problems) is a very transferable skill, it's about acquiring logic and methodology, understanding machines, cutting down complex problems. Those skills are good and useful.
Alternative. If you stop working on it what else will you do? If it's playing video games or watching tik tok, learn programming! If it's learning something else, will it bring you more value? Is it over a topic that will keep its value within the current trends? That being said, knowledge does not have to be applicable, learning for fun, is a good thing!
The capacity of technology is very hard to predict (the future is).
Disappearance of programmers. Right now, you have a lot of programmers, they are not out of job. At the current capacities, they will not be out of job ; to compare with art, I think that pure illustration is a done game. Most prominent people stating that programmers will disappear have vested interests in the AI hype.
Replacement vs enhancement. Right now any programmer is probably using AI. The thing it has impacted is problem solving and information retrieval. See https://www.reddit.com/r/programming/comments/1592s82/the_fa...
Side knowledge. Programming (and sometimes solving technical problems) is a very transferable skill, it's about acquiring logic and methodology, understanding machines, cutting down complex problems. Those skills are good and useful.
Alternative. If you stop working on it what else will you do? If it's playing video games or watching tik tok, learn programming! If it's learning something else, will it bring you more value? Is it over a topic that will keep its value within the current trends? That being said, knowledge does not have to be applicable, learning for fun, is a good thing!
At the beginning of the 1980s, shareware was everywhere. It seemed that there was an application to do anything you needed, ready to run, and only asking for $5-$50 in return for its continued use.
Thus, I was ready to swear on a stack of Bibles that there was NO MONEY WHAT-SO-EVER to be made writing computer software. I wanted to be a programmer, but I hedged my bets starting to earn an EE degree in college. Remember, this was right as Bill Gates rose to be one of the richest people on the planet.
So now, it seems there is no future in software, as there didn't seem to be then. It's an illusion. Software is complex, and thus has very complex requirements. Those requirements can't be specified... they can only be discovered. Your job is (as it always has been) to match the problem to the available tools.
In electrical engineering circles, this is called impedance matching, and it's a black art. In that world, there are only 2 independent values to worry about, resistance and reactance, but in software, there are an almost unlimited set of variables, so it's always going to be quite complex, and require you or someone like you, to make those optimizations.
Thus, I was ready to swear on a stack of Bibles that there was NO MONEY WHAT-SO-EVER to be made writing computer software. I wanted to be a programmer, but I hedged my bets starting to earn an EE degree in college. Remember, this was right as Bill Gates rose to be one of the richest people on the planet.
So now, it seems there is no future in software, as there didn't seem to be then. It's an illusion. Software is complex, and thus has very complex requirements. Those requirements can't be specified... they can only be discovered. Your job is (as it always has been) to match the problem to the available tools.
In electrical engineering circles, this is called impedance matching, and it's a black art. In that world, there are only 2 independent values to worry about, resistance and reactance, but in software, there are an almost unlimited set of variables, so it's always going to be quite complex, and require you or someone like you, to make those optimizations.
Within five or ten years (maybe a little longer for some physical jobs), AI and robotics will probably be able to do ALL current jobs. So you should tell those people they also need a new plan.
I think the best idea is to learn how to leverage AI and/or robots to create some product or service. AI and robot labor will be cheap within a few years.
But learning programming including how to program AI is a good path still.
I think the best idea is to learn how to leverage AI and/or robots to create some product or service. AI and robot labor will be cheap within a few years.
But learning programming including how to program AI is a good path still.
Replace every occurrence of AI with "outsourcing to India" in what your friends and brother told you. Then you will have been told the exact same thing everyone in the 2000's was told.
As long as someone has to be held accountable when something goes wrong, or has to translate human needs into software solutions, humans will need to be in the loop.
The better you are at programming, the better able you are to use AI to produce working software. It becomes a tool that expands what you're capable of doing.
Learning to program well without AI will also train you to be better at thinking and problem solving.
If you want to learn how to program, all of the above are in your favor. Just avoid using AI to learn since you won't be able to tell when it's hallucinating. Only use AI for things you already understand and can verify yourself.
The better you are at programming, the better able you are to use AI to produce working software. It becomes a tool that expands what you're capable of doing.
Learning to program well without AI will also train you to be better at thinking and problem solving.
If you want to learn how to program, all of the above are in your favor. Just avoid using AI to learn since you won't be able to tell when it's hallucinating. Only use AI for things you already understand and can verify yourself.
I'm assuming your big brother doesn't program also. Learning programming is still a very good skill and will benefit you even if AI becomes more prolific.
The usefulness of a conductor becomes greater the larger the orchestra becomes, not less.
The usefulness of a conductor becomes greater the larger the orchestra becomes, not less.
Learning coding is still valuable for the same reason knowing basic arithmetic is useful. You technically don't need to know how to add, subtract, multiply and divide since a calculator can do it better, but even so it's still an incredibly useful skill. I think coding is the same way.
Now, is software engineering still a viable long term career option? Honestly, probably not (to be fair, I think most white collar jobs will be significantly automated away anyways). But if it's something you're genuinely interested in, I'd say you should still give it a shot.
Now, is software engineering still a viable long term career option? Honestly, probably not (to be fair, I think most white collar jobs will be significantly automated away anyways). But if it's something you're genuinely interested in, I'd say you should still give it a shot.
There are a few layers to this problem. They're not entirely wrong about developments in AI-assisted coding – it is progressing quickly. But for now, LLMs doing "any software engineer job" is still far off. At least in the way people like your brother mean it: that some non-technical person can just tell an AI "build <this or that> for me," and the AI handles it end-to-end.
Until we reach that stage (if ever), technical people will still be needed in the loop – to check that the AI's work is correct, integrates properly with the rest of the system, and meets the requirements. And this leads to the main role of any engineer: figuring out the requirements. In other words, determining what exactly needs to be built.
People, in general, aren’t great at describing things in a precise, unambiguous way – and non-technical people even less so. I don't see this part of the equation being automated anytime soon.
As long as you enjoy figuring things out, solving technical problems, architecting software systems, and understanding software at a deep level, I think there will be work for you for the foreseeable future – no matter how "AI-assisted" it becomes.
Until we reach that stage (if ever), technical people will still be needed in the loop – to check that the AI's work is correct, integrates properly with the rest of the system, and meets the requirements. And this leads to the main role of any engineer: figuring out the requirements. In other words, determining what exactly needs to be built.
People, in general, aren’t great at describing things in a precise, unambiguous way – and non-technical people even less so. I don't see this part of the equation being automated anytime soon.
As long as you enjoy figuring things out, solving technical problems, architecting software systems, and understanding software at a deep level, I think there will be work for you for the foreseeable future – no matter how "AI-assisted" it becomes.
Don’t worry, your friends have no idea what they are talking about.
As long as you enjoy solving software engineering problems (not just copy pasting lines of code from sof) — you can have a great career.
No serious software engineer with at least 5 years of good hands on experience is worried about it.
Gpt etc is fundamentally not “intelligence” in any valuable sense. It is nothing but advanced and powerful auto-complete for your code.
It will never _replace_ a real engineer.
As long as you enjoy solving software engineering problems (not just copy pasting lines of code from sof) — you can have a great career.
No serious software engineer with at least 5 years of good hands on experience is worried about it.
Gpt etc is fundamentally not “intelligence” in any valuable sense. It is nothing but advanced and powerful auto-complete for your code.
It will never _replace_ a real engineer.
I think that some of the change in outlook for programmers has to do with the difference between now and ten years ago.
Ten years ago you could go to code school for 6 months, do a mediocre job and land an entry level job making 90k a year, and so the market was flooded with people who weren’t serious about programming.
So “should I learn to code” use to be similar to “should I learn to ride a bike” or “should I learn to read”. Like yeah, if you could learn to ride a bike and make $90k a year, you should do it.
But I think you need to ask it like you’re asking “should I learn medicine” or “should I learn electrical engineering”.
It’s a discipline, and one that takes commitment and constantly relearning. The old school, I am an instrument that turns instructions into code is over. But if you can master coding, and all of the soft skills that come with figuring out what needs to be done?
Then yeah, you should learn programming.
(This is all about career advice. Coding is fun! If you want to learn coding, learn coding)
Ten years ago you could go to code school for 6 months, do a mediocre job and land an entry level job making 90k a year, and so the market was flooded with people who weren’t serious about programming.
So “should I learn to code” use to be similar to “should I learn to ride a bike” or “should I learn to read”. Like yeah, if you could learn to ride a bike and make $90k a year, you should do it.
But I think you need to ask it like you’re asking “should I learn medicine” or “should I learn electrical engineering”.
It’s a discipline, and one that takes commitment and constantly relearning. The old school, I am an instrument that turns instructions into code is over. But if you can master coding, and all of the soft skills that come with figuring out what needs to be done?
Then yeah, you should learn programming.
(This is all about career advice. Coding is fun! If you want to learn coding, learn coding)
There will still be some dev jobs here and there, but not like before, and the ones that remain are probably gonna be like AI shepherds that oversee a bunch of AIs doing the work that you would've done five years ago.
As someone new to the field, I don't think you'll get many job offers. The entry level positions all seem dead, and will probably only get deader as AI gets better. It seems to me like career suicide to be learning programming now. IMHO only. In 30 years of programming, I'd never seen the field this bad.
I don't know how the industry expects to make new seniors without a pipeline for juniors, but you don't have to make that your problem too.
If I were you in your shoes, I'd learn something harder to automate (the trades, etc.). Robotics doesn't advance as quickly as software AI, so those jobs probably won't be replaced as quickly as programmers.
As someone new to the field, I don't think you'll get many job offers. The entry level positions all seem dead, and will probably only get deader as AI gets better. It seems to me like career suicide to be learning programming now. IMHO only. In 30 years of programming, I'd never seen the field this bad.
I don't know how the industry expects to make new seniors without a pipeline for juniors, but you don't have to make that your problem too.
If I were you in your shoes, I'd learn something harder to automate (the trades, etc.). Robotics doesn't advance as quickly as software AI, so those jobs probably won't be replaced as quickly as programmers.
Believe it or not, coding is the easy part. It's solving problems that gets you paid, and learning to program moves you in that direction if you're interested in learning how to solve tech related problems.
Just learn to program if you are interested in it. Tech skills in general are a good investment right now. And if AI does get good enough to replace your programming job, it will soon replace all other jobs too.
At least right now it can't. It makes a lot of mistakes. It is useful to write the code and then you just need to fix the stuff it didn't do right.
you should actually be more interested because now you can use AI too to assist you in learning programming.
This, with assistant you can learn faster than ever.
I think that Primeagen answers your concerns fairly directly in this video[1]
TLDR; Barely useful AI costs $5 per "bug fix" and you have to supervise it, that's the lowest compute, cheapest models... the "good" stuff is going to cost thousands per task, there needs to be 10,000:1 improvements in hardware before things really impact the market. However... code churn generated by the AI is likely to increase the need for competent programmers to sort out the mess. ;-)
[1] https://www.youtube.com/watch?v=kjOxkPl3RGo
TLDR; Barely useful AI costs $5 per "bug fix" and you have to supervise it, that's the lowest compute, cheapest models... the "good" stuff is going to cost thousands per task, there needs to be 10,000:1 improvements in hardware before things really impact the market. However... code churn generated by the AI is likely to increase the need for competent programmers to sort out the mess. ;-)
[1] https://www.youtube.com/watch?v=kjOxkPl3RGo