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

Technical Deflation

benanderson.work

Post reply on HN
Print view View original

71–77 of 77 posts

Re: Technical Deflation

#71

Post by Analemma_ »

vrighter wrote: post #42
Aurornis wrote: post #33

Earlier quoted context omitted.

> only really faster and better if you don't use them for gaming, unfortunately. Upscaling and frame generation is not a better GPU, it's one with a band-aid applied to hide the fact that it actually did not get much faster. The generations gains haven’t been as great as past generations, but it’s getting silly to claim that GPUs aren’t getting faster for gaming. Intentionally ignoring frame generation and DLSS up sc…

"Guessing what a pixel's color might be were one to actually do the work and render it" is not the same as actually rendering it. No, upscaling doesn't count. Doing it for a whole screenful of pixels, for the majority of frames (with multi-frame generation) is even less of it.

Gamers don't buy GPUs to do rendering, they buy them to play games and make them look good; rendering is just an instrumental proxy to the actual goal of making their games look good. If RTX and DLSS are accomplishing that, who cares how it is done?

Re: Technical Deflation

#73

Post by vrighter »

raw_anon_1111 wrote: post #47
vrighter wrote: post #46

Earlier quoted context omitted.

So? The point is that humans do it much less often. Let's say there are 10 subtasks that need to be done. Let's say a human has 99% chance of getting each one of them right, by doing the proper testing etc. And let's say that the AI has a 95% chance of getting it right (being very generous here). 0.99^10 = a 90% chance of the human getting it to work properly. 0.95^10 = only a 60% chance. Almost a coin toss. Even wit…

So humans do it much less often yet we have 30 years of evidence to the contrary? Humans still can’t figure out how to write code not subject to sql injection after 25 years or how to write code and commit it to GitHub without exposing admin credentials

yes, over time there are a lot of cases. Doesn't change the fact that llms make them more often. You want to look at the rate, not the absolute number.

Re: Technical Deflation

#74

Post by raw_anon_1111 »

vrighter wrote: post #73
raw_anon_1111 wrote: post #47

Earlier quoted context omitted.

So humans do it much less often yet we have 30 years of evidence to the contrary? Humans still can’t figure out how to write code not subject to sql injection after 25 years or how to write code and commit it to GitHub without exposing admin credentials

yes, over time there are a lot of cases. Doesn't change the fact that llms make them more often. You want to look at the rate, not the absolute number.

And of course you have evidence of this?

Re: Technical Deflation

#75

Post by kuboble »

makapuf wrote: post #54

Did google search and stackoverlow and internet deflate programming ? Comparing with reading books you might have or need to order, read, or get from the library, you bet . There still are some interesting problems to tackle. Maybe more than before. So who knows.

Having lived through the internet, Google and stack overflow revolting I think there is a substantial difference with an LLM.

The previous made it easier to learn things, while LLMs do things for you.

As a result of Google and SO I was really bad in what they replaced.

As a student I memorized all the utilities I had to use frequently, but since the search revolution I keep googlung simplest stuff. "how to check if character is in string" every day because I don't have to remember it.

I might be an exception but I clearly got much worse in knowing things by heart because they are one search away.

Amd I think it's plausible that a person who will only ever work with LLM will never learn to actually code. And if they don't then what will they be able to reliably deliver.

Re: Technical Deflation

#76

Post by vrighter »

raw_anon_1111 wrote: post #74
vrighter wrote: post #73

Earlier quoted context omitted.

yes, over time there are a lot of cases. Doesn't change the fact that llms make them more often. You want to look at the rate, not the absolute number.

And of course you have evidence of this?

gestures around vaguely

Have you been following the homepage of this site lately?

Re: Technical Deflation

#77

Post by raw_anon_1111 »

vrighter wrote: post #76
raw_anon_1111 wrote: post #74

Earlier quoted context omitted.

And of course you have evidence of this?

gestures around vaguely Have you been following the homepage of this site lately?

Anecdotes are not data. Is there any data that an LLM produces less secure code than your average developer?
Post reply on HN