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

Testing GPT 4's code-writing capabilities with some real world problems

tylerglaiel.substack.com

Post reply on HN
Print view View original

581–590 of 677 posts

Re: Testing GPT 4's code-writing capabilities with some real world problems

#581

Post by JPws_Prntr_Fngr »

BurningFrog wrote: post #413
vages wrote: post #253

Earlier quoted context omitted.

If increased productivity equaled job loss, there would be two programmers alive today, doing the same job as the fewer than 10000 programmers using punch cards as we entered the year 1950. A lot of projects today are not even greenlit because they would be too expensive to make. For instance, there are a lot processes in almost every country that require you to file paper forms, even though we have had web forms and…

Something like 97% of all jobs have been eliminated since the Industrial Revolution started ~250 years ago. And instead of the 97% unemployment our intuition tell us would happen, we're instead 30x as productive. The same will happen when/if AI makes writing software far more effective.

> … instead of the 97% unemployment … we’re instead 30x as productive.

Economic productivity simply means output per worker. So, we could make 97% of the US population permanently unemployed tomorrow, and as long as the AI replacement labor yielded equal or greater proceeds to the remaining 3%, we’d see a colossal increase in “productivity.”

That would be in the interest of the 3% to do (or attempt), which is what makes the scenario so scary.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#582

Post by coldtea »

dimitrios1 wrote: post #553
coldtea wrote: post #431

Earlier quoted context omitted.

First, this is an argument by authority. One of the "world's best computer science educators" can still say all kinds of wrong stuff. Especially someone as opinionated as Dijkstra, with many controversial opinions other world leading computer science educators disagree with. Second, relevance and applicability is depending on context. On the context of this discussion, about practical programming in the trenches, wha…

This is an argument of "challenging those who have the hubris to think Harold's statement is irrelevant to stop thinking small minded", if I were to call it anything. Additionally, even if I concede it is an argument solely on authority, just because an argument contains a fallacious point doesn't make it untrue, otherwise that would be the fallacy fallacy.

>Additionally, even if I concede it is an argument solely on authority, just because an argument contains a fallacious point doesn't make it untrue

Notice how you also conveniently sidestepped my arguments about why it's also untrue (on top of containing a falacious point).

Re: Testing GPT 4's code-writing capabilities with some real world problems

#583

Post by quonn »

chrismsimpson wrote: post #213
eggsmediumrare wrote: post #197

Earlier quoted context omitted.

If the outcome is the same, does it matter?

Have you heard of the halting problem? For one, you’ll never be able prove the outcome is the same.

In general. This does not mean that it could not be proven for most or even all relevant cases.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#584

Post by mrandish »

I'm actually quite excited by what the examples in the article show – despite the fact they show GPT-4 can't replace a good dev solving somewhat tricky algorithm problems.

Reason: I'm a code hobbyist who glues various modules together that have been written by much better programmers than I am. My end goals are never more ambitious than doing pretty simple things which I'm doing mostly to amuse myself. My biggest time sucks turn out to be tracking down fairly simple syntax things that vary between different languages and frameworks I'm slapping together (because I rarely spent more than a couple hours working on any one thing, I never get super familiar with them).

Being a lousy coder with little desire to put in significant effort to improve just to make my personal hobby projects a little easier, a basic AI assist like this looks pretty useful to me.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#585

Post by oblio »

anonzzzies wrote: post #564
oblio wrote: post #352

Earlier quoted context omitted.

And the output? Are you allowed to use their code for your own projects or company?

Yes, we don’t all live in the US.

So you basically don't care about licensing.

That's ok for your use case but risky for anything open source of that might be sold around the world.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#586

Post by weatherlite »

gwoolhurme wrote: post #300

I guess my genuine question is to the people who are saying this is a big deal and it will take our jobs. I am a bit lucky to where at the moment I am working in a "novel" field. Lets say though for the sake of argument that AI does come for SWE jobs. To be honest? I don't know what to do in that case, I have no backup plan, not enough to retire. The country I've lived in for 9 years is still through a work visa (hop…

We all need to save up money and think of a plan B. If there is no problem, worst case you'll have a bunch of money saved and a plan B you won't be using.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#587

Post by napier »

danwee wrote: post #183

I want to see GPT-4 dealing with this situation: - they: we need a new basic POST endpoint - us: cool, what does the api contract look like? URL? Query params? Payload? Response? Status code? - they: Not sure. Third-party company XXQ will let you know the details. They will be the ones calling this new endpoint. But in essence it should be very simple: just grab whatever they pass and save it in our db - us: ok, cool…

Companies will have their own homegrown models trained on doc and data corpus stacks and fine tuned facets served by MS0AI GPTX and other cloud shoggoth gigacorps. Company A’s model will talk to company B’s model and they’ll figure out all of the above and complete the job in the time it takes you to blink and take a sip of your coffee. Bureaucratic grind lag and lossy communication fuzz may be some of the first casualties of the next few years, and I can scarcely begin to predict any further out than that.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#588

Post by travisjungroth »

blibble wrote: post #485
WinstonSmith84 wrote: post #475

Earlier quoted context omitted.

If you tried GPT4, you probably understood it's not about feeding all the code in the world. GPT4 analyzes and "understands" your code and will answer based on this. Clearly, it will read the variable names and make deductions based on this. It will actually read the comments, the function names and make decisions based on this. And it knows the rules of the language. I mean, I'm writing this because this is what I'v…

it is entirely about feeding it all the code in the world it doesn't understand anything, it doesn't make deductions it's a probability model, and I understand how it's implemented perfectly well, thank you

Even view it as a simple probability model, you don’t need a million Python repos. The massive amount of English text + 1,000 repos + your codebase is very powerful. You can see this because you can make up a language, give it some examples and it’s surprisingly good.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#589

Post by comfypotato »

JPws_Prntr_Fngr wrote: post #581
BurningFrog wrote: post #413

Earlier quoted context omitted.

Something like 97% of all jobs have been eliminated since the Industrial Revolution started ~250 years ago. And instead of the 97% unemployment our intuition tell us would happen, we're instead 30x as productive. The same will happen when/if AI makes writing software far more effective.

> … instead of the 97% unemployment … we’re instead 30x as productive. Economic productivity simply means output per worker. So, we could make 97% of the US population permanently unemployed tomorrow, and as long as the AI replacement labor yielded equal or greater proceeds to the remaining 3%, we’d see a colossal increase in “productivity.” That would be in the interest of the 3% to do (or attempt), which is what ma…

But that’s never happened in the past. Why should it happen now? The Industrial Revolution is used as an example because it was the biggest spike in productivity, but the general trend has been the same since the invention of tools. Agriculture, another perfect example, lead to others having time to produce things instead of hunting/gathering. It’s easy to grasp with the agriculture example that “unemployment” wasn’t even really a risk.

Sure, there will be niche sub-industries that will be erased by LLMs, and people who have those niche skills will suffer. This has always been the case with technological advances. And always, it has not disrupted the economy.

Re: Testing GPT 4's code-writing capabilities with some real world problems

#590

Post by reaperducer »

danwee wrote: post #183

I want to see GPT-4 dealing with this situation: - they: we need a new basic POST endpoint - us: cool, what does the api contract look like? URL? Query params? Payload? Response? Status code? - they: Not sure. Third-party company XXQ will let you know the details. They will be the ones calling this new endpoint. But in essence it should be very simple: just grab whatever they pass and save it in our db - us: ok, cool…

Oh, wow. Your API story just described how I spent eight months of 2022.

I thought I was going crazy.

Now I'm sad that this is just how it is in tech these days.

Post reply on HN