> Creating software is starting to resemble cooking. Your home-cooked software is exactly what you need, without extra fuss or cost. And each time you build something personal, you gain a deeper appreciation for the craft of creating software.
I think this is true, but it also actually requires you to want to make your own software. In cooking, most people are aware that they're physically capable of making a roast chicken with gravy and mashed potatoes.
Yet takeaway/delivery is thriving. It's not that the ability to cook is scarce, but the desire to do it. I think the same is (now) true with building software.
AI doesn't enable non programmer to make their own app. It made non tech managers wishfully believe they won't have to pay coders to spend days to write the code.
I feel like existing models and platforms still have a bit of work to do to make this easy for everyone, but overall I agree that this is the direction things are headed. I have very minimal programming experience, but I was able to put together a personal project manager that I'm actually using in my personal life and at work.
I posted about it recently: https://news.ycombinator.com/from?site=cgarduno1garduno.gith...
Here's the repo: https://github.com/cgarduno1garduno/SPOT
The thing that blew me away was that this took ~2 hours total time and I didn't touch the code at all - this was 100% done using Cursor. Similarly, I've been building other tools that I use at work and my team is starting to roll them into our standard workflows because they work well.
There are many instances where I struggle to get it to make what I want, but overall I've been really impressed by how much it has helped me and by how easy it has been to start building my own personal software.
I lean toward most people remaining generally apathetic. There are many DIY kits and guides for a great number of things that have done most of the hard work of designing, measuring, cutting, etc for you. My perception is that most people would still rather buy or use an already finished product.
Does the average person even use computers outside of desk workers at work now, anyway? Can you build and deploy an app with an iPhone/Android app?
I think that people should give the idea that the common man is a boiling vessel of ingenuity and wonder restrained only by not having to right tools (read: products) a rest. I know that there are cultural and socioeconomic incentives to believing this. So the idea persists, for now.
I just don’t buy the supposition that in ten years those incentives will still attract the common man.
I don't think most people want to customize their software based on how many years of the Linux desktop there has been.
> We need more builders, not fewer. Because building fosters understanding. And as more people start making personal software, the bar for what counts as “great software” will inevitably rise.
Love this take. After over a decade of software development I've gained much more appreciation for well-built/useful products.
Random tidbit, I made this anemometer thing it collects data every second I think I'm at 20 million rows now since it's been running for years, what do I do with this data? nothing
It's not even calibrated to speed or anything I just have an ADC connected to a DC motor and logging the values/mapped to milliwatts via resistor value
I also had a solar cell but the coolest thing with that is seeing that bell curve or solar maximum
On the topic of personal software, I keep making note taking apps. I've made chrome extensions, mobile apps, Android widget, desktop app, etc... with multiple databases, end of the day it's just grabbing text, though recently I did go advanced and add in drag-drop image to base64 support (lazy) at some point I'm trying to unify all this data into one place with a sync mechanism
I appreciate the sentiment but this is totally unrealistic. I've tried to use AI several times before to help me quickly build out some personal software side projects and it took me waaaaay longer every time. It would just confidently spit out code that looked ok at first glance but if you actually know what you are looking at you'd realize it wasn't correct at all. Or worse, if it was an area I didn't fully understand such as native mobile dev, I'd burn an hour or two before realizing the problem wasn't me.
Yes build personal software. But to act like someone can leverage AI to do all the donkey work without actually understanding it is just silly.
The saddest part about this is that children, back in the 70s, were making personal software using Smalltalk.
The reason we even need AI to build "personal software" is because building modern software is terrible. Building for the web is terrible.
If we had a modern OS with easy to use app building tools and a language as easy to learn as smalltalk, then anyone could make software, including children. You can add AI on top of that and it's still better because at least you can understand what the AI made.
I find it amusing how the idea of ordinary people programming is talked about like it was something new, only made possible today by Dissociated Press on steroids, and it was NOT the default in personal computing even into the 90s with things like QBasic, only to be quashed by vendors who saw more money in "information appliances", especially with the web and then mobile ascendant.
Yes, an app can be a home-cooked meal. Back in the day, we didn't feel the need to eat out all the time.
I think spreadsheets were sort of an earlier version of this. You wanted to track some information, or make a plan, or just lay out some things, and this canvas of cells allowed you to put whatever structure _you_ needed in on the fly, adjust as you realized you needed something else.
It gave you the freedom and reduced the friction of iteration to figure out how you needed to organize things to discover what you wanted.
But it brought many limitations that surprisingly (to me at least) still persist today. I think AI has the ability to give you the same kind of freedom and friction reduction to allow you to build what you need.
HyperCard on the Mac solved this problem on System 6 through Mac OS 9. It was very easy to build your own stacks and even build stand-alone apps. With HyperCard, MacWrite, and MacPaint or even ClarisWorks (alter AppleWorks) bundled with the hardware, pre-OS X Macs were fairly personal in both hardware and software right out of the box. And when Apple started to bundle Claris Emailer and Claris Organizer with Macs along side SoundJam (and later iTunes), you really didn't need much more for the average Performa user.
If AI was a real person, the author was licking its ass.
I haven't seen anyone with zero computer skill build apps but I've definitely started building and deploying apps outside my day job that I've been able to put together quickly with the aid of Claude artefacts or copilot. It's really exciting!
> Software can now adapt to you, not the other way around. Better yet, AI is making it possible for anyone, not just developers, to create single-use or custom applications.
Yeah, because that wouldn't be a customer service / troubleshooting nightmare.
> But within the next decade, millions of people will be able to create software. Designers, marketers, product mangers, and others will be the first.
Hahaha - clearly the author is young. All of this has happened before, and it will all happen again.
This is the only argument I've seen for why a second look at generative AI for (new) software developers might not be an unmitigated train-wreck…
…yet the author's points are completely invalidated by the lack of any explanation for how they intend to address the myriad of horrifying ethical lapses involved in how these models are developed and deployed—not to mention why most of the key influencers in this "AI" movement are deeply anti-craft and anti-humanist.
We in the pro-craft protest movement are happy to engage in good-faith conversations about which types of "AI" technologies might be good at in the future given satisfying resolutions to the host of ethical issues currently involved, but articles like this which seem shockingly unaware of this fierce debate simply come across as uninformed, and uninformative.
> Creating software is starting to resemble cooking.
I'm so tired of the relentless hype around AI. No normal non-technical person is going to build and be able to maintain their own software.
It's very easy to see where a lot of VC money is going: marketing.
AI writing code for you is a terrible way for a beginner to "make their own software."
We've already had solutions to this for years, it's called jamming a bunch of plugins on to a Wordpress site so it roughly does what the person wants it to do.
or perhaps remember Visual Basic?
Why does every god damn thing on the website need to be about how "AI will change everything"? You do realize we've had these tools since the 90s, right?
There's a reason only a small percentage of people are actually interested in making their own software. Most people just want to use the thing and be done with it, not tinker around with code.
How many people are making their own clothing? Clearly sewing machines are widely available, so they will surely create a revolution where everyone will make their own clothing! And yet - do you want to actually make your own clothing? No. You want to have time for other things, like making your own software.
Has anyone done experiments with people who have absolutely no programming or programming-like experience and see what they can come up with?
From my experiences, any output I got from "AI", required already some level of experience/understanding of the tooling or language to get something useful out of it.