Rediculous: software developers have been obsolete since 1957 when FORTRAN was developed. As it was described at the time: ‘Specialized computer “programmers” will no longer be needed as the computer may now be directly instructed by anyone through easy, English-like commands.’
"Ridiculous" was a joke, taking the fantastical claims about FORTRAN as true (thus it's "ridiculous" to imagine the problem going away in a decade when it has already been solved for over 60 years).
I first saw this claim 20 years after that (in the late 70s) and have seen it pretty regularly since.
FORTRAN did expand the set of people able to write code, but it still required as much rigor as before. And the claim has always been true in trivial cases: famously secretaries who "couldn't program" were writing Emacs macros at MIT in the 70s. Visicalc and now Excel allow people who don't believe they can program to write reasonably complex programs, but it quickly becomes unmanageable, undebuggable spaghetti.
IDEs have long been touted as programming-eliminators, but the code they generate has always been brittle and limited, and in the end...can do only trivial things.
Sure, at some point many scriptable or programmable objects may be composable by people without training, but we will need human programmers until machines themselves are at least as smart as humans.
FORTRAN did expand the set of people able to write code, but it still required as much rigor as before
How far this set of people should expand for software development to become like driving, which still requires rigor, but is accessible to almost everyone? Do you see this happening as a result of the latest AI breakthroughs, or are there any other fundamental challenges?
As the last 20 years of AI research has ignored semantics I can't see it helping in a significant way.
I suspect it will be a while after driving requires zero rigor (or even participation) before computers could program themselves.
Frankly, given how hard it is to explain to another human how to do something why should we expect smart computers to be easier to manage? Unless...we use a highly structured manner of issuing instructions...i.e. a programming language!
The fact that it didn't make software developers obsolete. Instead it created a whole new breed of 'specialized' software developers - ones that knew Fortran.
This is the fundamental problem with "they are going to automate our jobs!" thing people seemingly fear. Sure automation may make old jobs obsolete, but it also spawns new jobs in their place.
Before the invention of cars - there were no auto mechanics.
Before the invention of telephones - there were no telephone operators.
Before the invention of the internet - there were no web developers.
Before ATMs - there were surely more teller jobs in existence. But with ATMs, someone is now needed to go around and stock these ATMs with money periodically, as well as someone to repair them occasionally - both of which can be viewed as a new kind of "teller".
As so, software developers will never become obsolete. It's just the way in which we execute our jobs, which may become obsolete...
So, if a user says "create a website that does x, y, z", and describes in plain English, on a high level, how he wants this website to look like, and AI creates the website which does what the user wanted, does this mean this user is now a "software developer"?
Because the only requirement for the new breed of 'specialized' software developers might be they know English.
>does this mean this user is now a "software developer"?
Sure. He developed software - who cares what tools he used to do it? Is a farmer who farms via tractor not a farmer?
The OP gave an apt example in FORTRAN - as FORTRAN is literally something which automates writing (lower level) code for you. Are software developers who use FORTRAN not software developers?
>Because the only requirement for the new breed of 'specialized' software developers might be they know English.
You assume everyone knows 'English'. To my knowledge only ~20% of the world "knows" basic English - let alone the specific English required to instruct an AI to craft a website precisely to do what you want it to do.
Every time I see an article with a heading like this I just rename the headline to something like this (especially since the last few paragraphs of the article contradict the headline):
"Why clickbait journalism might be obsolete by 2030".
Some tasks are better done by tools now. I agree with that.
I believe that we still haven't figured out how much software we need. Or put another way: Every software breeds new software.
How did we get to low code platforms in the first place? Because of software. Who is using low code? Specialists. What can you build on top of platforms? More software. What established new ecosystems? Software.
Software is a tool. There is still virtual reality to come. And virtualization simply shows how much value software can generate.
Example: Virtual goods. Whoever thought that human beings need stuff that only exists in a game hosted in the cloud?
Excel? Coda, Notion, etc.
Maybe also the term developers itself changes. Remember the Frontend vs Backend discussion? This happens to Data Scientists and Python as well.
I believe that by 2030 there will be even more demand.
Oh man. It's like Visual Basic all over again! "It's so easy - just fire your entire department and have the secretary write the codes - look, you just drag and drop!"
I'm waiting for the predictions about how self-driving tech was going to eliminate all truck driver jobs by ... last month or so?... to come true first before I start worrying.
I'm curious if the author of the article ever created systems tailored to a specific business. Website WYSIWYGs that create "beautiful" websites have probably taken a chunk of "Web Designer" jobs. (Square Space as an example).
Automated testing tools are created to help the developer and possibly save company some money by not hiring a dedicated QA person.
This is definitely not the first time I heard this, but of those who are claiming this BS were non-developers.
Damn, every week same thing over and over again. I didn't scroll down the article and already can assume that may contain something about GPT-3/OpenAI.
> For example, if you ask a computer to print("I am so stupid") two-hundred times, it will do so without complaining, and complete the task in a fraction of a second. Ask a human, and you’ll need to wait for hours to get the job done…
In a minute I can do 10 `print("I am so stupid")`. So I would say 30 minutes is a more reasonable timeframe for a human.
Where all predictions misses the point is when they think technology is iterative and that we learn and improve from previous mistakes.
That is false for most of the cases. We keep reinventing the wheel and making the same mistakes, just in a slightly different and incompatible way. We have also been getting worst at establishing standards, which could be used as building blocks for utopia AI automated everything.
Software development is not going anywhere different.
there are still some major fortune 500 companies who can't even make their website work properly, let alone begin any kind of automation. I don't think we'll see major parts of the economy automated in our lifetime. Maybe small parts of an industry here and there but not most of the economy.
AI's first spoken words on second day at the job: "What in the hell goes on here anyway? Geeeez"
Also software development makes all intelligent actors into mumblers usually after year 7 ... so we can expect to hear, if we strain our ears, some more choice words from Captain SourceCode. Gonna be fun!