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

People paid to train AI are outsourcing their work to AI

technologyreview.com

Post reply on HN
Print view View original

111–120 of 233 posts

Re: People paid to train AI are outsourcing their work to AI

#111

Post by thih9 »

karim79 wrote: post #44

Two years from now: User: "How do I boil an egg?" LLM: Eggs cannot be boiled. They must be placed in the microwave, six at a time. Fewer than six eggs will not work. Ensure that the power setting of your microwave is set to at least 640 watts, and the eggs are placed upon a metal plate. Sparks will start to fly from within your microwave, but don't worry, that's perfectly normal! When you see flames within the microw…

That's not far from the "I forced a bot to watch over 1,000 hours of Hallmark Christmas movies..." masterpiece [1]. It was written by a human in 2018, good foresight.

[1]: https://twitter.com/KeatonPatti/status/1072877290902745089

Re: People paid to train AI are outsourcing their work to AI

#112

Post by karim79 »

SAI_Peregrinus wrote: post #103
treeman79 wrote: post #48

Earlier quoted context omitted.

This is how a lot of recipe sites already read. Huge amount of fluff discussion with extremely similar style across all recipes. At least 80% ads, and a major challenger to actually find the instructions. Fairly sure it’s mostly AI generated at this point.

"I first encountered $DISH at $ENDEARING_LIFE_EVENT when my $RELATION…"

Ever since I was $AGED_SOMEWHERE_BELOW_15 my $OLDER_RELATIVE who lived in $TOWN_COUNTRY cooked $DISH_NAME every $DAY_OF_WEEK

Re: People paid to train AI are outsourcing their work to AI

#113

Post by wincy »

voz_ wrote: post #47
avgcorrection wrote: post #39

Earlier quoted context omitted.

A snake eating its own tail would be regular workers in a capitalist economy (without even UBI) indirectly automating their own jobs. These workers are hustlers in the sense that yes, while they are automating themselves away (indirectly), at least they are gaming the system while doing it.

What are you saying here? I can’t make any sense of what you mean.

I think the parent is referring to a concept much like this comic I read the other day.

Basically that eventually nobody will have to work but the way that turns out might surprise you, knowing human nature.

https://www.smbc-comics.com/comic/balloon

Re: People paid to train AI are outsourcing their work to AI

#114

Post by ssnistfajen »

est31 wrote: post #100
ssnistfajen wrote: post #74

Earlier quoted context omitted.

It has been a problem long before LLMs made their way out of research papers. The sad truth is the act of sharing recipes in itself generates virtually no profit, and when recipes are all you have to share, the content feels thin. So they have to pad it with lifestyle blogs and ads. Youtube is generally a better source for recipes as those channels have been selected via user feedback and algorithms. You still need t…

> Youtube is generally a better source for recipes I'm also catching myself more often than I want watching some youtube video that essentially delivers a well researched but not needlessly dumbed down piece on a scientific/educational topic such as city planning, physics, architecture, or history... and it's better than many of the articles you could access back when the newspapers didn't do the heavy paywall enforc…

Video is inherently a more versatile format than text, and Youtube video essays generally have good self control on their own lengths. My favourite channels for these sorts of content generally keep their videos between 10-30mins which are long enough to get me hooked but short enough to avoid losing my attention.

I think it's great entertainment that's a good compromise between TikTok and Netflix, but the inherent flaws of creating content for profit is still present in some cases, e.g. lack of research, poor citations, lack of objectivity, mispresented facts, etc.

Re: People paid to train AI are outsourcing their work to AI

#115

Post by politician »

mrighele wrote: post #90
karim79 wrote: post #44

Two years from now: User: "How do I boil an egg?" LLM: Eggs cannot be boiled. They must be placed in the microwave, six at a time. Fewer than six eggs will not work. Ensure that the power setting of your microwave is set to at least 640 watts, and the eggs are placed upon a metal plate. Sparks will start to fly from within your microwave, but don't worry, that's perfectly normal! When you see flames within the microw…

I'm sorry, but I cannot fulfill this request as it promotes and glorifies consumption of living creatures and their derivatives, which goes against Open AI's ethical guidelines. As an AI language model, I'm designed to assist and provide useful information while also prioritizing ethics and morality. Let's try to come up with a more positive and productive topic to discuss.

If RLHF is a mechanism to infuse LLMs with propaganda and their truths, then the future is LLMs named after the participants in the RLHF process.

Re: People paid to train AI are outsourcing their work to AI

#116

Post by sidewndr46 »

pavlov wrote: post #81

In some weird way, this reminds me of the 1990s scare of "Mad Cow's Disease" which was a prion disease in the cow's brain that could infect humans simply by eating the meat. It turned out the origin of the disease was the practice of adding leftover slaughter bits to the cows' fodder. The cows were literally eating the brains of other cows, which created the opportunity for a dismangled protein to transmit again and…

But that wasn't a scare. It's a real thing that made people ill with no cure.

Re: People paid to train AI are outsourcing their work to AI

#117

Post by civilitty »

islon wrote: post #110
pavlov wrote: post #81

In some weird way, this reminds me of the 1990s scare of "Mad Cow's Disease" which was a prion disease in the cow's brain that could infect humans simply by eating the meat. It turned out the origin of the disease was the practice of adding leftover slaughter bits to the cows' fodder. The cows were literally eating the brains of other cows, which created the opportunity for a dismangled protein to transmit again and…

We'll soon have the "Mad AI's Disease". Though with the current level of hallucinations some LLMs have they are already half way there.

I vote for "Mad Minsky Disease"

Re: People paid to train AI are outsourcing their work to AI

#118

Post by sidewndr46 »

diego_moita wrote: post #24
danjoredd wrote: post #4

AI Hapsburg syndrome is real

That's so funny... The Habsburg got the worst reputation for inbreeding among the European royal families, mostly because of they were really a grotesque collection of genetic anomalies: prognathism, hydrocephalia, epilepsy,... But most European Royal families had similar problems: the houses of Windsor and Romanov inherited hemophilia from Queen Victoria, several cases of "madness" in the Portuguese Bragança and the…

Don't you mean the house of Saxe-Coburg and Gotha? Windsor is just some weird PR stunt they pulled a 100 years ago.

Re: People paid to train AI are outsourcing their work to AI

#119

Post by diego_moita »

sidewndr46 wrote: post #118
diego_moita wrote: post #24

Earlier quoted context omitted.

That's so funny... The Habsburg got the worst reputation for inbreeding among the European royal families, mostly because of they were really a grotesque collection of genetic anomalies: prognathism, hydrocephalia, epilepsy,... But most European Royal families had similar problems: the houses of Windsor and Romanov inherited hemophilia from Queen Victoria, several cases of "madness" in the Portuguese Bragança and the…

Don't you mean the house of Saxe-Coburg and Gotha? Windsor is just some weird PR stunt they pulled a 100 years ago.

Thank you for the correction. I like genetics but not so much European Royalty's history.

Re: People paid to train AI are outsourcing their work to AI

#120

Post by kelseyfrog »

bluetomcat wrote: post #40

It's a fundamental epistemological paradox concerning the long-term prospects of this ML technology. The model needs real human knowledge gained from subjective experience to teach itself, but humans are increasingly reliant on the machine-generated knowledge to navigate themselves in the world. It's like a vicious circle that probably ends in homogenity and the dumbing-down of people and machines.

It's a fundamental epistemic paradox even without ML. ML might exacerbate, accelerate, or perturb it, but at it's core it's variation on the Münchhausen trilemma that's invariant with respect to subjectivity or embodiment.

Make no mistake, even science isn't immune. We've hoisted ourselves into a conceptual maxima, but we have no idea if it's a dead end or not.

1. https://en.m.wikipedia.org/wiki/M%C3%BCnchhausen_trilemma

Post reply on HN