Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI-enhanced development makes me more ambitious with my projects

simonwillison.net

Post reply on HN
Print view View original

481–490 of 519 posts

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

#481

Post by ummonk »

SanderNL wrote: post #126
ummonk wrote: post #27

Earlier quoted context omitted.

Our experience coding from scratch will have about as much validity as the experience of people who coded in assembly has today. It'll be a specialized skill with its own niche but without much value in the broader tech industry.

If that was all they were doing, sure.. but if they wrote language parsers, understood computing from the nand-gate up and worked their way up and down the stack over-and-over.. these types of guys will blow past all of us. The understanding you get from going up and down stacks is ridiculous.

I mean I've worked up and down most of the stack. The understanding is great. It still doesn't give me much of a competitive advantage over the average "full-stack" (laughable term given how narrow a part of the stack it encompasses) engineer when we're all working at FAANG / generic SaaS startup churning out agile code commits for generic CRUD web app.

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

#482

Post by The_Colonel »

CharlieDigital wrote: post #206
speedylight wrote: post #127

Earlier quoted context omitted.

I honestly have issue with using ChatGPT to write medical software. I don’t know what your exact process is like but I hope you’re giving the code it generates extra scrutiny to make sure it really does what you put in the prompt. It kinda feels like the judge who used ChatGPT to determine whether to deny or grant bail.

Medical software has to go through a validation process. As long as that validation process is still as rigorous, I don't see much difference.

> Medical software has to go through a validation process.

Guess what will power the validation process?

Think about those efficiency gains.

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

#483

Post by manicennui »

xxswagmasterxx wrote: post #441
tinyprojects wrote: post #188

100% this. I just used GPT-4 to help build an analytics dashboard using ChartJS. There's so many settings in ChartJS, it would have taken me a week to StackOverflow / Google how to get my charts how I'd like them - it took me a day with GPT-4. I could just ask it anything and it would help no problem. Any buggy code it produced, I'd just copy and paste the error message and it would provide a fix. The day before I bu…

Imo the chartjs docs are not very good. Or i'm just used to rust's docs.rs, where the whole api surface is documented. I couldn't find something like this for chartjs.

You mean like this?

https://www.chartjs.org/docs/latest/api/

I just started looking at the docs a couple minutes ago to understand why the OP claims that it would have taken a week to accomplish something with this library. I don't get it.

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

#484

Post by manicennui »

tinyprojects wrote: post #188

100% this. I just used GPT-4 to help build an analytics dashboard using ChartJS. There's so many settings in ChartJS, it would have taken me a week to StackOverflow / Google how to get my charts how I'd like them - it took me a day with GPT-4. I could just ask it anything and it would help no problem. Any buggy code it produced, I'd just copy and paste the error message and it would provide a fix. The day before I bu…

How is it possible that any task could take a week or more with this library? It looks extremely straightforward and the documentation contains lots of examples.

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

#485

Post by manicennui »

asim wrote: post #61

Every generation faces this sort of issue which is learning skills for the current technology which then get displaced in 20-30 years. I think there's always a worry we're going to lose some knowledge in the process but equally the productivity gain seems to be 10x and it leads to dealing with higher level problems. Very few people are kernel hacking now as a percentage of the overall engineering community. I guess t…

We have been told that computers will do our (software engineers) jobs for decades now. It has never come to fruition, and this is no different.

What we are seeing is that the number of software engineers has increased dramatically and the average skill level has plummeted. It is not normal for a competent software engineer to need to reference Stack Overflow a dozen times per day every single day.

The other thing that this is bringing to the forefront is just how atrocious the libraries are in certain ecosystems (e.g. JS/TS).

If a tool that can write incorrect functions/methods makes you substantially more productive, you were never very valuable in the first place, and "AI" likely isn't the only thing threatening your job security.

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

#486

Post by tracerbulletx »

jsty wrote: post #148
JimmyRuska wrote: post #38

Earlier quoted context omitted.

Having spent a lot of time with GPT-4 recently, it's possible the new generation will not have the same issues we did. For example they can write some code and say, "adjust for best practice". Then "explain why needs to be " Oh this is called , give me an example of it. Give me another example. Give me another example. Give me use cases where this would be better. Give me a short story where it was done my way but it…

> Then "explain why needs to be " It would be nice, but until there's a breakthrough on having LLMs / other models generate, maintain and walk the user through a representation of their reasoning, these explanations are likely no more reliable than a random StackOverflow answer - except without the benefit of other users scoring / picking holes in the answer

No further breakthroughs are necessary, if you're experienced enough you have enough general context to tell when the advice is good and it's helping you understand a new subject and it absolutely is great at walking you through understanding something new in a broader domain you're already familiar enough with to detect when it's lost.

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

#487

Post by ddren »

arcatech wrote: post #463
ddren wrote: post #347

Earlier quoted context omitted.

I would trust ChatGPT code about as much as I trust the code produced by any human. All the Therac-25 code was written by a human, so what is the argument here exactly? At least when you tell ChatGPT that its code is wrong it agrees and tries to fix. Ok, it usually fails at fixing it, but it doesn't refuse to acknowledge that there is a problem at all, unlike the case of the Therac-25. I like to think that it is not…

In general we would like developers/engineers to know as much as possible about the things they're engineering. ChatGPT-based development encourages the opposite.

So because ChatGPT exists now, less experienced programmers will be hired to developed critical software under the assumption that they can use ChatGPT to fill the gaps in their knowledge?

Even in that case, I would argue that is entirely a problem of the process, and should be fixed at that level. An experienced programmer doesn't become any less experienced just because they use ChatGPT.

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

#488

Post by roncesvalles »

ThePhysicist wrote: post #30

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…

This is sarcasm, right?

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

#489

Post by beauzero »

elorant wrote: post #453
ren_engineer wrote: post #410

Earlier quoted context omitted.

Am I the only one that's been disappointed with GPT for coding? It's good at using common libraries and frameworks but for stuff using more niche libraries it fails miserably almost every time, to the extent that by the time I've debugged and written prompts to try and get it to do it properly I've spent more time than needed still good for handling a lot of grunge work and really useful for doing the stuff where I'm…

I'm on the same boat. Every time I asked it to provide C# code it gives me back crap. Nine out of 10 times it’s using non-existing libraries. The same thing happened the other day when I asked an SQL Query. I’m still on ChatGPT 3.5 though, perhaps 4 is way better.

Started with ChatGPT, migrated to BingChat, now onto Bard...added Copilot at the same time as I started using ChatGPT. Settled on Bard (has gotten increasingly better towards the end of the week) and Copilot

Using Copilot as a better intellisense...but don't use it for big templates. Bard to find/summarize stuff I figure is probably in stack/SO somewhere.

Boilerplate I think I have seen a 35% increase in speed. Individual edge cases (like porting Azure AD v1/Identity 2.0 to Azure AD v2/Identity 2.0 authentication) maybe 10-15% improvement. My day to day is C#.

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

#490

Post by tjr »

hgsgm wrote: post #472
HeyLaughingBoy wrote: post #404

Earlier quoted context omitted.

Going meta for a bit: before you can use a tool to produce medical device software, that tool must be qualified for use. I'd really like to see what the qualification process for ChatGPT would look like.

What is the qualification for using StackOverflow or a library book? What's the qualification for the keyboard that might introduce errors (hello, MacBook) or the monitor that might render improperly?

Not answering for medical industry, but answering for the similar realm of aerospace systems:

One big question is, does the proposed software tool assist a human engineer, or does it replace a human engineer?

If a tool replaces a human -- the phrase used often is "takes a human out of the loop" -- then that tool is subject to intense scrutiny.

For example, it would be useful to have a tool that evaluates the output of an avionics box and compares the output to expected results, to automatically prepare a test passed/failed log. Well, this would amount to replacing a human who would otherwise have been monitoring the avionics box and recording test results. So the tool has to be verified that it works correctly, in the specific operating environment (including things like operating system version, computer hardware type, etc.)

So what about ChatGPT? One big hurdle is that, given the same input, ChatGPT will not necessarily provide the same output. There's really no way to verify its accuracy in a repeatable way. Thus I doubt that it would ever become a tool that replaces a human in aerospace engineering.

How about using it then to assist an aerospace engineer? Depending on the assistance, this should not necessarily be materially different than getting help from StackOverflow.

Post reply on HN