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

GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

artisana.ai

Post reply on HN
Print view View original

101–110 of 151 posts

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#101

Post by lumost »

troops_h8r wrote: post #7

I don't think I see enough discussion about what this means for privacy. There was some protection in the fact that it was prohibitively expensive to get someone to listen to every single one of our phonecalls/read all our emails/etc. Worrying that this will no longer be the case.

The top use case I've been hearing is in legal discovery. Law firms used to play games with diligence by disclosing TBs of email and making it cost prohibitive to find relevant emails. This task would normally require a $60-100/hr paralegal or lawyer.

GPT-4 can do that task for fractions of a penny per email now. It doesn't have to be perfect if its competing with nothing. I expect we'll see similar shops for any other high cost paper/trail business.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#102

Post by snotrockets »

two_in_one wrote: post #98

>This breakthrough saved the researchers over $500,000 and 20,000 hours of human labor. BTW, this is interesting. There is a lot of noise about AI carbon footprint. Now imagine how much humans would eat and fart for 20.000 work hours. It's about 10 man/years. Assuming 8h / 5d / 50 weeks schedule.

The best time to delete that comment was before you wrote it. The second best time is now.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#103

Post by snotrockets »

adastra22 wrote: post #62
876978095789789 wrote: post #50

Great to see this tech and the money invested in it being used to take low-paying jobs away from people with limited options, instead of something like drug discovery or cancer biology.

Better get used to it. That's basically tech's entire value proposition.

It's also useful to launder biases that would be immoral, if not outright illegal, if humans would admit to employing.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#104

Post by thatsadude »

SoftTalker wrote: post #69
dragonwriter wrote: post #30

Earlier quoted context omitted.

Shotspotter is a thing and it already has been doing that for years (both on its own and on law enforcement request.)

Shotspotter isn't an AI though is it? I thought it was just triangulation of gunshot locations using microphones and synchronized clocks?

That's a layman explanation. ShotSpotter is likely a passive radar system. In recent years, you can combine signal processing and supervised learn (neuralnet) to get better direction-of-arrival estimations.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#105

Post by nr2x »

RC_ITR wrote: post #77
nr2x wrote: post #56

Earlier quoted context omitted.

How sure are you about that? The basic theories have been around since the 70s, have been proven at scale in the last decade, and the NSA has more data and compute than anybody else. I’d be shocked if they aren’t very far along in solving many problems.

This isn’t really a “throw money at it” problem like the government is good at. Take drones for example. The government got really good at those because they made them jet-powered (lol) and blew a bunch of money on server-grade FPGA’s in each one of them. You can’t really just buy a lot of GPUs to make an LLM work, you need iterative development of architecture and training methods. Like maybe the government invented…

I wouldn’t rule out NSA being ahead of the curve, but you have a good point re: GPUs. Likely another factor in the CHIPS act.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#106

Post by nr2x »

stevenhuang wrote: post #92
nr2x wrote: post #56

Earlier quoted context omitted.

How sure are you about that? The basic theories have been around since the 70s, have been proven at scale in the last decade, and the NSA has more data and compute than anybody else. I’d be shocked if they aren’t very far along in solving many problems.

Pretty sure. Even the Manhattan project had nuclear research going on in public universities at the time. Nothing of the sort here for the attention mechanism which underpins LLMs we know today. Fundamental research isn't something you just throw money at and acquire. All we had back then were cleverbot and other expert systems.

More my point is they have as big a research budget as a corporate lab.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#107

Post by HeavyFeather »

two_in_one wrote: post #98

>This breakthrough saved the researchers over $500,000 and 20,000 hours of human labor. BTW, this is interesting. There is a lot of noise about AI carbon footprint. Now imagine how much humans would eat and fart for 20.000 work hours. It's about 10 man/years. Assuming 8h / 5d / 50 weeks schedule.

Indeed time to eliminate all those people I guess. /s

I don’t think you can compare people’s carbon footprint because those people will exist regardless of jobs.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#109

Post by ilaksh »

ftxbro wrote: post #95
ilaksh wrote: post #85

Earlier quoted context omitted.

GPT4 is already better than most if not all humans in some metrics. I think it's a hard sell to say that nothing better than human is allowed. But the existential concern comes from supposed exponential takeoff. To me it should be easy to convince people that something 10 or 100 times faster or smarter than humans should not be allowed. Weirdly you don't see people talking about regulating autonomy which is also part…

Probably we both think similar things, but I disagree with one thing. I think it's weird that you say "I think it's a hard sell to say that nothing better than human is allowed." in combination with "To me it should be easy to convince people that something 10 or 100 times faster or smarter than humans should not be allowed." To me, those situations are way too close to each other (human level vs. 10x 'smarter') to b…

We already have the human level and people seem unconvinced. And it's been shown to be safe as far as existential risk.

The easily convincing argument for me is imagining the planet populated by aliens that move extremely slowly. This makes more sense when it's closer to 1/100th speed. https://www.lesswrong.com/posts/5wMcKNAwB6X4mp9og/that-alien...

Although having actually read that, it's not convincing in this context because the AI in the story is already fantastically performant.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#110

Post by flandish »

bequanna wrote: post #86
hnaouesteuho wrote: post #58

From reading the paper, GPT-4 also outperformed the researchers themselves in many categories, despite the researchers being the ones who created the dataset being used to perform the comparison. In other words, the metrics are biased in the researchers’ favor — so GPT-4 would have beat them even more often (probably a majority of the time based on the numbers), if someone else had created the guidelines and golden l…

With the current unskilled labor shortage driving wage increases which pushes inflation up, this seems to be arriving just in the time.

Just a small point of order: there is no such thing as “unskilled labor” -

All labor is skilled labor. Even breaking rocks.

Post reply on HN