legit concern I hear is its environmental impact.
Commuting my fat ass to a climate controlled office, out to lunch, back home, parking spaces, highway lane miles, fuel, periodic vehicle replacements… that all has environmental impacts too, if I can do my job in half the time, that’s a big win for the environment.
I have been doing this stuff for over 40 years, the tools get faster and the ecosystems get more complex.
What would be really nice is a return to simplicity, using the fast tools to make simple stuff fast-squared, but nobody seems to want that.
If you ask something longer than 20 lines, there’s a very high probability that it won’t work on the 15th round of corrections.
Try Claude by Anthropic. I noticed Copilot and Google getting hung up much faster than Claude.
Also, I find that if you encourage a good architecture, like a formalized system of variables with Atomic / Mutexed access and getter/setter functions, that seems to give a project more legs than letting the AI work out fiddly access protection schemes one by one.
I got one up around 500 lines before it started falling apart when trying to add new features. That was a mix of Rust and HTML, total source file size was around 14kB, with what I might call a “normal amount” of comments in the code.
Once my business is in a more profitable place I’ll bring someone on to fix up the code
AKA: technical debt. I actually approve of this approach when you’re testing the market and don’t have any paying customers. Where it gets ugly is when customers start placing trust in your product, trust that might be costly if your code fails, and management doesn’t budget the resources to actually fix up the code. I was very glad to leave the place that was doing this…
The problem is when people assume you can actually build an entire software/service architecture of any complexity just through vibe coding.
Welcome to CEO handling 101. It’s an art, a very soft skill, and not for the faint of heart. I worked for a mid sized (50 employee) company once where I’d “speak truth to power” in our weekly meeting, get shot down rather enthusiastically by the CEO during the meeting, then after I and the rest of R&D left his office, he’d go out to production and have them start implementing all the concepts of my pitch - as his own ideas, naturally.
That’s a great tip: having it review the security of code that an earlier context generated.
I plan on having it write unit tests, or at least try to…
You need to be able to read it to understand that it’s going a little off the rails.
At least 2/3 of the time I spend with AI coding is getting it to compile without errors - that’s more than a little off the rails, but it’s also much more helpful when you finally do get to a working example that you can look at, instead of beating your own head against the Stack Exchange archives hoping for inspiration, let it try for you.
I have been building various things with AI coding tools for a month or so now. I rate the various engines on how far I can take them before they get hopelessly lost, unable to correct their own errors. For the best tools this seems to come after about 50 to 70 iterations of asking for small feature additions or error corrections, weaker tools (like Copilot) hit these infinite loops of fixing their errors with other errors much faster.
It’s a good limit, because after 2-3 hours of AI interactive development, I can then spend 4-6 hours going through the resulting code - cleaning it up and understanding how it works. I suspect if AI were taking me farther, like 100-150 iterations, it would probably take me more like 15-20 hours to unravel the various things it comes up with - kind of a point of diminishing returns.
Bottom line: think of your project in terms of microservices. AI is pretty good at microservices. As long as the individual services are each robust in their delivery of the required functions, you’re in good shape.
If it ever becomes “mystery meat,” it’s time to recode by hand.
AI coding is actually a very powerful tool, almost like a light saber. Do you notice how many amputations and artificial limbs there are in that galaxy far far away?
Because: for $20 per month to the AI company, you can output poor code much much faster.
I had a period of about 10 years where I bounced from company to company fixing non-programmers’ code so that it could actually be used in commercial products that brought in revenue.