Live data from Hacker News

GPT-4 is phenomenal at Code

github.com

121–130 of 221 posts

Re: GPT-4 is phenomenal at Code

#121

Are software engineers going to go the way of human “computers”? I know GPT-4 won’t put us all out of work, but I worry that something could in a number of decades I can count on one hand. Of course, it would be gradual. I am excited, disappointed, and as a young person who hopes to be a SWE for a long time, somewhat afraid. Then again, it could be argued that this merely means GPT4 is a compiler…

> Are software engineers going to go the way of human “computers”?

Anything with digital IO will.

Re: GPT-4 is phenomenal at Code

#122
post #50

> is phenomenal at Code Can it - attend bullshit meetings - lick my manager's boot - tell them that their ideas are good but in the masochistic, self-deprecating way they like to hear it from me in Not yet? Darn it, guess my job is safe for now.

Yes, it can. - Automated relations, whether with your employer or a customer, are trivial - These relations can be fine-tuned for exactly the level of syncophancy your boss prefers - So yes, it can tell your boss their ideas are good in precisely the masochistic, self-deprecating way, etc. Your post is not longer than a system prompt. Think it through.

But it will have zero credibility, so no.

Re: GPT-4 is phenomenal at Code

#123
post #84

Earlier quoted context omitted.

I don't think this is a good test. Every CS graduate student has to write a lambda-calculus parser. There must be thousands of implementations on the web. It really is not strange that GPT-4 can reproduce this.

Let's invent a better test then. What could be an example task that isn't widely available on the internet already? I'm having a hard time coming up with anything that isn't reproduced in many places.

Take something from advent of code and change it up a bit seems like a good test

Re: GPT-4 is phenomenal at Code

#124
post #8

Well, I think I just retired. Even knowing this was coming, it's still a bit unnerving. Sad too, but it's for the best. Let the machines program the machines. Schmidhuber's Gödelmachine. The automatic scientist .

Just the opposite, this is letting me pick up more work than ever. The nature of that continues to shift as rote memorization becomes less important, but that was already the case thanks to earlier forms of code completion. It may be that breadth is becoming more valuable than depth. The engineering design process is more important than ever, along with soft skills and an eye for what kind of work matters.

>this is letting me pick up more work than ever

Is this a desirable outcome, to be just a meat robot middleman for the machines doing the actual work? I hope this will lead to a total rethinking of how our relationship to work and the why of it all.

Re: GPT-4 is phenomenal at Code

#125
post #15
post #4

Any concerns about sending code directly to a third party? I’m very hesitant to use any tool that is sending my assets to the cloud to be used who knows how.

The large tech company I work at explicitly told us not to upload code or any private company info to ChatGPT or similar services. I assume other companies have the same policy.

I think companies are getting very receptive to uploading code with the new 30-day policy. We have been able to negotiate this for financial firms. Reach out and we might be able to make something happen.

Re: GPT-4 is phenomenal at Code

#126
post #61

Is it? According to OpenAI's paper GPT-4 GPT-4 (no vision) GPT-3.5 Leetcode (easy) 31 / 41 31 / 41 12 / 41 Leetcode (medium) 21 / 80 21 / 80 8 / 80 Leetcode (hard) 3 / 45 3 / 45 0 / 45 https://cdn.openai.com/papers/gpt-4.pdf Table 1; page 5. So it's better than GPT-3.5, but still pretty pathetic at hard Leetcode problems. If your programming job is closer to leetcode easy problems you might be in trouble, but for the…

people are impressed because you never need to do leetcode hard problems in any business anywhere in your career to make a few billion dollars. I would fire anybody using company time trying to reinvent some academic search thing instead of just using the hashmap.

Reimplementing quicksort in a language the rest of the company doesn't use in a part of the software that no one ever touches is fireable, sure, but I suspect that on the way to ChatGPT there were more than a couple "Leedcode hard"-grade problems that OpenAI engineers and PhDs solved because the library for that didn't exist before.

Re: GPT-4 is phenomenal at Code

#127
post #36
post #8

Well, I think I just retired. Even knowing this was coming, it's still a bit unnerving. Sad too, but it's for the best. Let the machines program the machines. Schmidhuber's Gödelmachine. The automatic scientist .

All this does is change how you go about getting computers to work for you. Someone off the street still needs to know how to express themselves correctly to get a result that's not crap. And they don't know what they don't know. So this thread will be filled with people freaking out about the ends of their jobs. When really it just changes how their jobs work. The nailgun didn't retire carpenters. They just got more…

Pre-industrialisation: 80% of population are farmers

Post-industrialisation: There's going to be a big squeeze in the field (pun not intended).

Re: GPT-4 is phenomenal at Code

#128

> is phenomenal at Code Can it - attend bullshit meetings - lick my manager's boot - tell them that their ideas are good but in the masochistic, self-deprecating way they like to hear it from me in Not yet? Darn it, guess my job is safe for now.

It can lick boot perpetually and without rest. Many managers, however, will still prefer the real thing.

Re: GPT-4 is phenomenal at Code

#130
post #104
post #66

Earlier quoted context omitted.

I remember it being vastly overhyped about how artists were now obsolete, but it really hasn't happened. People have quickly became very attuned to the "AI art style"--humans are excellent at pattern recognition, after all--and if anything, it's made people appreciate how important an artists' individual style and attention to detail is.

It's much too soon to say, no? AI art is going to keep getting better quickly, and DALLE-2 only came out six months ago.

Progress can come to a screeching halt at any point. We saw this with self-driving cars previously: the AI true believers were certain self-driving would be everywhere within 5 years... in 2015. Now the idea of a self-driving car has mostly been abandoned.

Let's get at least one company, at least one, replacing its graphic design department with DALL-E, or its development team with GPT, before calling an entire profession obsolete.

Post reply on HN