Live data from Hacker News

AI-enhanced development makes me more ambitious with my projects

simonwillison.net

221–230 of 519 posts

Re: AI-enhanced development makes me more ambitious with my projects

#221
post #203

Earlier quoted context omitted.

> Are people writing code then just blindly shipping it to production? Of course they do. Ship fast, break things, look cool, cash out, leave someone else to fix the mess you have created, it's the new black.

And AI makes this...worse?

Yes, it speeds it up considerably. More heaps of messy software produced much faster.

Re: AI-enhanced development makes me more ambitious with my projects

#222

Earlier quoted context omitted.

I can read the code and verify it. Sometimes there are small mistakes but overall the architecture and approach are mostly correct. It won't produce 100 % production-grade code, but that's not expected for a PoC. And personally I also get stuff subtly wrong at times, I just iterate and test to find these cases and fix them. So I guess we're not fully replacable yet.

You can't read API calls and know if they're real, or if the function names are plausible but non existent. This is the problem with GPT. It creates great sounding possible APIs, but has no idea which APIs actually exist.

> You can't read API calls and know if they're real, or if the function names are plausible but non existent

Why not? And if I miss them, my editor, compiler or general testing should pick that up.

Re: AI-enhanced development makes me more ambitious with my projects

#223

Yesterday I asked chatgpt to compute the probability that two words of five letters both began with the same three letters and it could not solve it. It gave me about 10 different results but all were wrong, it seems the number five mislead the algorithm. Another question was computing the variance of the sum of 60 dices but omitting the results where the sum was greater than 200, the reasoning was incorrect in all c…

I've seen stuff like that too. I asked ChatGPT a tricky geometry question and it got a ratio reversed and ended up producing wrong answer.

I have also tried to use ChatGPT like Google search at the beginning by asking short, curt questions, and got underwhelming results.

The trick is not to try to trick ChatGPT but pretend that you are an idiot and it is a very sophisticated colleague who just needs step by step details. Then you can also generate some of the seemingly amazing results people are getting. For example yesterday I used it to scrape physician information from my insurance website, followed by their Google ratings and so on.

Re: AI-enhanced development makes me more ambitious with my projects

#224
post #98

Earlier quoted context omitted.

How do you know it is even close to correct? When I’ve asked it for code, it has been subtly wrong, or worse, correct but missing important edge cases that it doesn’t even seem to be aware of.

Make it write a test suite that you can manually verify/expand as necessary, just like code written by humans. I don't understand this 'but how do you know it works' argument. How do you know the code you wrote yourself works? I hope you don't rely on your superior programming skills to just get it right the first time?

You build a correctness proof in your head by thinking through all the details. That what takes most of the time, and the AI doesn’t relieve you from that. Of course one doesn’t always get everything right, but testing alone is no substitute for deductive thinking about the code’s correctness. And it’s also less fun having to do that for someone else’s code.

Re: AI-enhanced development makes me more ambitious with my projects

#225

Honestly, I keep secret some of the amazing uses of ChatGPT. I feel selfish but also I feel as if everyone is respsible for what they get out of it. We have many people and the media spending hours and tons of prompts trying to create funny or eyebrow raising headlines. I just keep my head down an marvel at the growing array if uses, many complex, that chatgpt can perform. It has been making me tons more money for ju…

Same. I'm almost inclined to include prompt engineering as a skill in my CV next time I'm out looking for jobs.

Re: AI-enhanced development makes me more ambitious with my projects

#227
post #154

Earlier quoted context omitted.

You are sort of ignoring the bias in your own AI extremist prediction. People who have decided to go all in in learning an interpreted language like Perl or even PHP, had to continuously also learn react, js, ruby,... and gazillions of other languages and frameworks to remain in business over the last two decades. While people who invested in learning C, Java or SQL have as much healthy opportunities to pursue today…

There are no perl jobs.

There are a considerable number of Perl jobs out there - just not many "greenfield" perl jobs.

Most of them are maintaining existing codebases (not a fun job), and some are "help port this legacy codebase to not perl".

I'd not discount Perl. Its incredibly fast to develop services in, stable as fuck, runs anywhere, performant (relatively), flexible, and if you were a startup looking to get a product out the door, just offer decent TC to a couple of old hands you can ship something viable fast.

Re: AI-enhanced development makes me more ambitious with my projects

#228

Earlier quoted context omitted.

Electron is still an abomination. I've gone from hating PHP for most of my career, to now having a grudging appreciation for it. Its a clusterfuck, but it works really well, and deployment barely gets simpler than "untar into webroot, done".

There is still required setup for things like PHP, like what reads from the webroot and serves the requests. Other things, like binaries, can be almost completely self-contained and do everything, and then you just move the binary to the server and run it, and you get the full environment needed, no Apache with mod_php required (or what the kids use nowadays)

I'll grant you that.

I also forgot briefly what a complete fucking nightmare instrumenting/debugging PHP (on say, a LAMP stack) can be - which is the inevitable result when PHP decides to behave in "interesting" ways.

Re: AI-enhanced development makes me more ambitious with my projects

#229

What I really love is the unbounded optimism of the AI. How many colleagues could you ask "Can you write me a PoC for a horizontally scalable, probabilistic database in Rust, using Raft-based consensus?" that would then just think for a second and go "Certainly! Let's start by ...". These people exist, but they're one in a thousand (or million?). ChatGPT basically does that for almost every prompt you throw at it. So…

Most of mine would probably, correctly?, advise C++ and Paxos for that use case but each to their own!

Depends how jr your team is I suppose.

Re: AI-enhanced development makes me more ambitious with my projects

#230

Earlier quoted context omitted.

It’s crazy to think about. We are nearing the end of the age where people who have been programming for “20 years this year” and self-taught probably learned to program from books. In the next few years, people who have been “programming for 20 years” might start to include YouTube learners and boot campers. As someone who learned to program from two gigantic tomes of C/C+ with a Borland Compiler, there was no copy/p…

I’ve been programming for 10 years, and as you’d expect know a lot of people with a few more or less years programming. I think all the “kids these days” gripes only apply to people’s first few years. You can argue they’re formative, but most professional programmers will quickly need to evolve beyond “take message literally, ask online for help, paste error message into google/SO” to be good at their jobs. I don’t k…

Hey! I still do that!

But it's true that after a point, the only errors left are the ones that can only be resolved by reading the documentation and experience (only 4 DELETE per hour, and you throw me an undocumented 503 error? Really?), and llm can't really do it for you.

Post reply on HN