> if I can’t convince you, maybe this article from 2017 where Andrej Karpathy (director of AI at Tesla/OpenAI) will.
Yeah... uhh not to throw too much shade, but isn't a car driving itself simpler than writing software? If programmers are going away then does that include the ones working on self-driving cars? Hell, wouldn't this displace all engineers, lawyers, doctors, etc. in general? Are the businesses of the future going to be run by even more heartless robot CEOs? Why isn't that also discussed? Why are these stories always about "programming"?
If AI replaces programmers and engineers, does that mean Tesla will become the world's first business without any humans working there? ;)
Writing code is "easier" than most other things, especially driving, because you get unlimited do-overs (up to practical limits). If the code doesn't compile, tweak it and try again. If the car doesn't stop for a pedestrian... uh oh!
I'll wager the opposite because even modern AI sucks at anything other than language interpretation. AI is absolutely awesome for history recall, but ask it to innovate and you will find where the limits lie.
Also I've yet to run across a single AI that can successfully implement a new feature in a 10+ year old code base that consists of a hundreds of thousands of lines of code and a gazillion different dependencies.
Believe me, I've tried. It would make my job MUCH easier.
Well, with AI's ability to complete more tasks companies will get smaller anyway. No programmers means there isn't a need for managers for those programmers. It means there aren't separate teams with different priorities. Someone "at the top" will speak to the computer, the change will be implemented and taken care of automatically. There will be no organizational politics to navigate.
> By comparison, software engineering is entirely digital, exponentially scalable, and always behind schedule. It’s the perfect target for a hostile takeover.
So, as an engineer, it may be more inherently job-secure to work with a software engineering market that is required to be somewhat analog, is inherently less scalable, and generally easy to keep on schedule (likely because that's not where the real problems are for that market, or they don't really care about measurements of time in proportion to traditional engineering environments).
In a context like that, any prospective AI takeover will automatically be awkward, a risky business bet by its very nature, and inherently resisted by stakeholders for good reasons. Grass is greener for AI elsewhere, and will be for decades at a minimum.
You'd want to connect your services further up the chain toward where people with hearts and loneliness and all the other human foibles dwell. People who need people.
And you'd want to take on the burden of dealing with AI on behalf of those people.
Yep. This article is clearly low-effort. The very few examples are like this one: """imagine you inherited a slow-performing code base written in Python. Your IDE suggests it be translated to Rust. You click yes and redeploy, and the app is suddenly 10X faster"""
If it's a complete redesign, then why ten only, if it's typing and inline-ing, then the author is very late to the party.
> the full-stack generalist is gone, as well as the product team and mobile developer. Niche fields like video gaming and medical software are now under threat. Developers transition to other engineering fields like robotics and biotech. All of this commences the age of the super SME (subject matter expert) — non-technical users with deep business knowledge who build businesses directly through AI.
So far in my career, developing software to solve problems for customers I’ve had to interface with SME’s. As described usually someone who is non-technical but not always. Eventually after a few years on the project I know the same amount as the SME but usually more. I know more because I also understand the software system. So I’m not a software developer. Nor a programmer. Im a builder. I’ll pivot to being that SME who builds businesses through AI.
Others of us may go into other engineering fields if they feel like it.
Also high technology is usually not solvable by AI. ATM it only knows what it’s been told.
> Long-held competitions between frameworks will be decided by which is the most AI-friendly. For example, imagine one could safely upgrade a legacy JS codebase to React with a single button click, but doing the same migration to Vue would take a week. Even if there are viable reasons to choose Vue, ultimately frameworks that adapt quickest to AI integration will prevail.
Might be true in some types of software development. A business where software isn’t the core sure. If the business needs a CRUD app sure. But software is a large domain that encompasses multiple types of business. There is a reason why certain technologies are used.
> For example, imagine you inherited a slow-performing code base written in Python. Your IDE suggests it be translated to Rust. You click “yes” and redeploy, and the app is suddenly 10X faster.
Ridiculous statement. Python can be fast. This is very much dependent on what you’re doing. Also good luck getting scipy functionality in Rust. And if a refactor like this was this easy and that critical to business it probably would have already been done.
> Coding in an IDE will die off. In fact, the “code” is nothing more than an implementation detail, forgotten more with each passing day. The sole focus of these products becomes the human-AI interaction.
Already true today. Replace human-AI interaction with senior-junior developer interaction. As I’ve moved up in my org code is nothing more than implementation details. I’m focused on selecting technologies and ensuring interfaces between large software components developed by different teams working together to solve a problem. I may dive into code here/there and give my 2¢ on how something should be done. Looking forward to AI being my new junior devs. And my junior devs becoming seniors.
> Software is now unrecognizable. Business rules are maintained by AI-powered SMEs. AI-native standards are globalized. Code no longer resides in static repositories — it is ephemeral and dynamic.
A good idea and I can’t wait. Someone should build a quick POC where you write a .d.ts file have a “compiler” use GPT to implement all the declarations.
Cybersecurity is going to grow. AI will make mistakes.
> Imagine your future self owning a digital business. You create and maintain it through an AI-powered design platform.
As the article implies, if anyone can use AI tools and no need for a technical person, would businesses exist? Would everyone have a business then?
Here’s an equally valid thought since we’re just speculating here:
Me: Create me an iPad app to watch YouTube without ads.
AI: deploy to mdmglr’s iPad?
me: Yes
me: actually, create season 6 of Silicon Valley and stream it to my iPad
AI: now streaming on your iPad
So it’s not going to be just developers, it will be writers, actors, etc.
If so, they'll call us "technical analysts" or similar, as I've yet to work at a company where the PMs and UX people full understand the logic of what they are trying to make. "Programmers" seem to fill this gap mostly.
Not only do they fill this knowledge gap, but the gap is a massive chasm. Implementation details are not just figuring out how to get something to work in a bubble without regard for the rest of a system, but finding good compromises based on very imprecise and highly product-specific requirements. AI can just get something equivalent to a poem to barely compile when real world systems are a series of heavy tomes that contain nothing at all what the PMs think they do.
It's naive to think the capitalist class won't try to automate us all. Whether they will succeed is highly speculative. At the current pace we'll know it rather sooner than later.
28 comments
[ 2.4 ms ] story [ 65.8 ms ] threadThe people who say
- new technology can cause dramatic paradigm shifts - this is new technology - therefore this will cause a dramatic paradigm shifts
are as bad as the ones who say - there many examples of new technology which did not "work" - this is new technology - therefore this will work
I remember when Segway was going to change the nature of transportation.
LLMs are showing a lot of promise, but trying to predict the future is a coin toss..
Yeah... uhh not to throw too much shade, but isn't a car driving itself simpler than writing software? If programmers are going away then does that include the ones working on self-driving cars? Hell, wouldn't this displace all engineers, lawyers, doctors, etc. in general? Are the businesses of the future going to be run by even more heartless robot CEOs? Why isn't that also discussed? Why are these stories always about "programming"?
If AI replaces programmers and engineers, does that mean Tesla will become the world's first business without any humans working there? ;)
Also I've yet to run across a single AI that can successfully implement a new feature in a 10+ year old code base that consists of a hundreds of thousands of lines of code and a gazillion different dependencies.
Believe me, I've tried. It would make my job MUCH easier.
So, as an engineer, it may be more inherently job-secure to work with a software engineering market that is required to be somewhat analog, is inherently less scalable, and generally easy to keep on schedule (likely because that's not where the real problems are for that market, or they don't really care about measurements of time in proportion to traditional engineering environments).
In a context like that, any prospective AI takeover will automatically be awkward, a risky business bet by its very nature, and inherently resisted by stakeholders for good reasons. Grass is greener for AI elsewhere, and will be for decades at a minimum.
You'd want to connect your services further up the chain toward where people with hearts and loneliness and all the other human foibles dwell. People who need people.
And you'd want to take on the burden of dealing with AI on behalf of those people.
(A lot of engineers already do this...)
New features, bug fixes, etc. Who does that? Just input all the exist code in the prompt and have the AI do whatever is needed.
I have few things to pick on this time:
> Job Loss Prediction: 5%
Has anyone lost their job yet to ChatGPT?
> the full-stack generalist is gone, as well as the product team and mobile developer. Niche fields like video gaming and medical software are now under threat. Developers transition to other engineering fields like robotics and biotech. All of this commences the age of the super SME (subject matter expert) — non-technical users with deep business knowledge who build businesses directly through AI.
So far in my career, developing software to solve problems for customers I’ve had to interface with SME’s. As described usually someone who is non-technical but not always. Eventually after a few years on the project I know the same amount as the SME but usually more. I know more because I also understand the software system. So I’m not a software developer. Nor a programmer. Im a builder. I’ll pivot to being that SME who builds businesses through AI.
Others of us may go into other engineering fields if they feel like it.
Also high technology is usually not solvable by AI. ATM it only knows what it’s been told.
> Long-held competitions between frameworks will be decided by which is the most AI-friendly. For example, imagine one could safely upgrade a legacy JS codebase to React with a single button click, but doing the same migration to Vue would take a week. Even if there are viable reasons to choose Vue, ultimately frameworks that adapt quickest to AI integration will prevail.
Might be true in some types of software development. A business where software isn’t the core sure. If the business needs a CRUD app sure. But software is a large domain that encompasses multiple types of business. There is a reason why certain technologies are used.
> For example, imagine you inherited a slow-performing code base written in Python. Your IDE suggests it be translated to Rust. You click “yes” and redeploy, and the app is suddenly 10X faster.
Ridiculous statement. Python can be fast. This is very much dependent on what you’re doing. Also good luck getting scipy functionality in Rust. And if a refactor like this was this easy and that critical to business it probably would have already been done.
> Coding in an IDE will die off. In fact, the “code” is nothing more than an implementation detail, forgotten more with each passing day. The sole focus of these products becomes the human-AI interaction.
Already true today. Replace human-AI interaction with senior-junior developer interaction. As I’ve moved up in my org code is nothing more than implementation details. I’m focused on selecting technologies and ensuring interfaces between large software components developed by different teams working together to solve a problem. I may dive into code here/there and give my 2¢ on how something should be done. Looking forward to AI being my new junior devs. And my junior devs becoming seniors.
> Software is now unrecognizable. Business rules are maintained by AI-powered SMEs. AI-native standards are globalized. Code no longer resides in static repositories — it is ephemeral and dynamic.
A good idea and I can’t wait. Someone should build a quick POC where you write a .d.ts file have a “compiler” use GPT to implement all the declarations.
Cybersecurity is going to grow. AI will make mistakes.
> Imagine your future self owning a digital business. You create and maintain it through an AI-powered design platform.
As the article implies, if anyone can use AI tools and no need for a technical person, would businesses exist? Would everyone have a business then?
Here’s an equally valid thought since we’re just speculating here:
Me: Create me an iPad app to watch YouTube without ads. AI: deploy to mdmglr’s iPad? me: Yes me: actually, create season 6 of Silicon Valley and stream it to my iPad AI: now streaming on your iPad
So it’s not going to be just developers, it will be writers, actors, etc.
There is a convergence...
Seniors etc will see their wages slashed through labour competition.
App and website clones will abound.
It won't replace programmers entirely, although it will diminish numbers significantly.