If other engineers are getting the productivity boost that I am from AI, then we are just scratching the surface of it's effect on the economy.
And if they aren't, then they will be soon enough.
And if they aren't, then they will be soon enough.
There should be some amazing new end-user-facing software, or features in existing software, or reduced amounts of bugs in software, any day now...?
https://youtube.com/@groove-tronic
Now you can't say nobody has anything to show.
This does real-time DSP (time stretching, effects, analog emulation), in a language I didn't know when I started (C++), that I started a couple of months ago.
I could not have created this without AI, cannot make progress without AI. When I use up my Pro account limits, I'm done for the day - I'm too slow without it for continuing to make sense.
And this is why untold billions are sunk into AI. Creates a dependency on their services and removes agency from end users.
Edit: yes I’m aware it adds agency in the sense that some people can do things they couldn’t do before… as long as that’s permitted/surveilles, which is where the agency is lost.
I've don't see how I've lost agency. It's not like it's made me a worse coder: getting so much faster with AI hasn't made me slower without it.
> > cannot make progress without AI. When I use up my Pro account limits, I'm done for the day
Aside from the contradictory statements, I did address your points in my edit. I get the tradeoff. As long as whatever you do is in line with the providers of the services, and as long as you're ok with the surveillance, then sure... It's a tradeoff that MUST be considered and be aware of..
You can get stuff done without these.
Local models exist.
Why that might not translate to an increase in software quality and feature count is left as an exercise to the weary senior reader.
With AI, refactoring is cheap and safe.
Ah, I misunderstood. I guess I don't care much if AI was used to create the code that enabled a revolutionary feature vs. AI being the revolutionary feature.
But I do in terms of "is it a industry-changing advance in software development, the profession" which is what this thread seemed to be about.
https://youtube.com/@groove-tronic
I'm disabled (which is why I've been forced into starting my own venture) and can only work a few hours a day, but I'm still more productive than I have been in 25 years, even when I had a team of engineers and a designer working for me.
The code is also cleaner - because refactoring is cheap now.
And I'm working in a language (C++) that I had barely touched when I started.
The 0-to-1 learning curve reduction is definitely very real, but the open question for me is what's next.
It's a cross platform PXE server. You just run it (root-less, no config) and the other computers on your LAN can boot up via PXE and (via netboot.xyz and iPXE) automatically download a Linux installer
The tool itself and the website both started their life as extremely functional 1-shots from GPT-5. pxehost was in a ChatGPT chat, and pxehost.com began as a 1-shot in a Codex CLI.
To me it's really cool that something like pxehost exists, but the fact that it began life as a fully working prototype from a single ChatGPT response is insane.
Not suggesting it wasn’t useful, or that it’s not remarkably convenient. Just easy to forget just how much went into providing it, in terms of the human labor involved with its training data, financial investments, and raw resources. In the broader context, it’s an unbelievably inefficient way to get to where we got. But as long we are here, I guess we should enjoy it.
That claim would have much more force if you could point to a repo. Otherwise, it just seems like blind bias.
Given their access to models, tooling and insane funding/talent - they still suck in standard software engineer like the rest of us, if not more because of the pace.
All the AI integrations so far have been a joke, PoC level quality software. Talk to me when AI helps them rebuild core products into something more impressive.
I don’t think over the last 5-8 years there has been a shortage of code-slingers, as evidenced by all the tech layoffs. Using LLMs to generate more code does not equal productivity. There’s that famous story about the -2000 LOC commit, etc.
From observation and induction, it seems very easy to contribute net-negative value in a software development position, and delivering long-term net positive value really requires a lot of discipline from the developers, or a lot of wrangling from non-technical members.
PS: Discipline here does not mean "being picky about what things to accept" it means being picky about what things you make up. From observation, a lot of the most valuable work is super fucking boring, and it takes discipline not to make up a more intellectually stimulating task.
I could not have attempted to create the application I have created without it.
As far as money-making - I wager it's helped me a great deal more in regards to preparation for marketing and sales, since I barely knew where to start with that. But I don't have the tangible proof yet, since I'm just starting that process up.
If you can create an application easily with AI, then someone else can also create that application easily with AI. This means the value is no longer in that application merely existing.
To me, this means that the value, i.e. the things people will pay for, in software will increasingly come from branding on the low end and from the applications that AI can't yet create on the high end.