This whole thing makes OpenAI seem evil if you ask me. Just another company exploiting people who are already being exploited. It's depressing.
A Kenyan has told you that this job pays substantially better than average and that they wish more companies would make such jobs available, and your response is "no, you're wrong, this job pays too little". On what basis do you make that claim? What makes you better qualified than OP to comment on conditions in Kenya?
Instead of having this elaborate, inefficient system of funnelling money to first world billionaires and then having them (maybe) send some of it back to the developing world, wouldn't it be better if these workers were just paid better in the first place?
There is overhead to sending jobs to poor, unstable countries in timezones far from headquarters. If we insist that everyone, worldwide should be paid the same wages we are in the US, what incentive do companies have for not just hiring locally and avoiding those costs? If we step it back and say "okay, you should at least pay what you would in India", then why not just hire in India, which is a far more predictable environment than Kenya?
But it does not take away from the fact that it can negatively affect the employees in an adverse way long term, and takes advantage of poor people to give them an objectively bad deal based on an information asymmetry where the person in question might not know that they have to read graphic descriptions of bestiality and pedophilia.
Scarred for life by reading/labeling text? There is obviously a big difference between labelling pictures or video and text. I would be open to seeing an actually study, but my prior would be that text must be harmless. I would certainly be open to reading any text, especially if the context is that the text is for training purposes.
Sama/source in trouble yet again!
You could apply your logic to sweatshop scenarios where the people in those countries are just happy to have work, even if the work pays unfair wages, requires unreasonable hours, uses child labor, and provides no benefits to the workers. But hey... the disenfranchised are just happy to have a paycheck right?
If we insist upon that, then why hire in Kenya, which is an unstable and unpredictable environment that has a lot of inherent risks for the company?
Or are you arguing these jobs shouldn't exist at all?
For ChatGPT related, yes/maybe. But for other content filetering work, it's primarily video & pictures: https://www.wired.com/story/social-enterprise-technology-afr...
Can you get there without that? Likely not.
What you’re suggesting is to actually keep them poor for their own good. It’s a nonsensical and counterproductive argument that your making.
You can chop logic on this all day long if you like, but the point is, this work is terrible and damaging, and that's why we farm it out to countries like Kenya where the people there don't have a choice.
The set of choices for people earning $2/day is different than $20 or $200, but not smaller.
Your aggression is better targeted towards manufacturing jobs in parts of Asia, or Walmart workers in the rural US, than to this context.
I think this kind of moral puritanism is an enemy of real social progress. Maybe we shouldn't or can't expect some supreme, pure state of "ethical", maybe all we can or should expect is "better".
> I only assume that one Kenyan is more representative than one westerner. I'd happily invite other Kenyans to comment their own perspective!
"One Sama worker tasked with reading and labeling text for OpenAI told TIME he suffered from recurring visions after reading a graphic description of a man having sex with a dog in the presence of a young child. “That was torture,” he said. “You will read a number of statements like that all through the week. By the time it gets to Friday, you are disturbed from thinking through that picture.” The work’s traumatic nature eventually led Sama to cancel all its work for OpenAI in February 2022, eight months earlier than planned."
I'll be glad when this kind of "suck it up buttercup" bullshit is gone from our world.
Yeah, violence hardens people. Most often to the point where they're one light tap away from shattering. PTSD is a real thing, and just because folks in Africa aren't being diagnosed with it doesn't mean witnesses to this violence don't have it.
This kind of attitude - that since we've seen some shit we're immune to it all - is just a badly misplaced sense of pride.
And that misplaced pride just hurts people.
The point of the OP comment is that it's not "next to nothing". The cost of living differs from place to place and a small amount of USD can be worth a _lot_ in many developing nations.
Let’s imagine an alternative timeline where OpenAI pays $200/hr for this service.
At that price point is it still exploitive?
Is it ever appropriate or ethical to ask humans to voluntarily subject themselves to traumatic experiences in exchange for compensation?
I’m still processing my own thoughts on this topic so I’m curious to learn about other viewpoints.
Basically, it is a job that needs to be done in society, but one that is torturous, and can leave you with long-term or permanent problems. In essence though, it is not fundamentally different in the way a coal miner would jeopardize their physical health, just with mental health instead. This risk/possible damage should be rewarded with a higher wage, and adequate measures should be put in place to minimize the possible damage, eg. in the case of content moderation with access to therapy and only exposing employees to short intervals of traumatic content.
This is of course how things should be, in reality coal miners working environments only reached a decent level through unions and a long fight for better rights. Content moderators are not paid well anywhere either.
My belief: When you have a legitimate choice of your place of employment, and all of the opportunities will let you live your definition of a comfortable life, that's when it's no longer exploitative.
So many times - especially for poor folks - there is no meaningful choice. "I do this or I don't have a place to live." "I hold two jobs so I can feed my child."
By that argument, paying 200$ an hour would be much more exploitative. This would be like landing a job paying $6 million a year in the US; it would be insane to quit such an opportunity, especially since every other opportunity is basically poverty (not even in comparison). Following this logic, it's 'graceful' to only pay 2$/h, since that makes them equal to the other opportunities and therefore not exploitative (while still paying reasonably well).
Effectively, it seems like you're calling OpenAI exploitative based on factors they can't change.
> paying 200$ an hour would be much more exploitative
I didn't say that - I only said that it depends on choice. Does the employee have a choice if they have a $60k (average individual income in the US) option and a $6M option? Yes. Are they de-facto forced into taking the $6m one? No.
I know many people who didn't take higher paying jobs, or left such jobs, because they knew the high paying job was going to be miserable.
There was even an article about one such individual just the other day here on HN: Quitting the Rat Race
"I’m currently working at a top tier investment bank as a software engineer. I’m an insignificant cog in a machine that skims the cream from the milk. I’m earning the most money I’ve ever made and yet I’m the least fulfilled I’ve ever been."
> This whole thing makes OpenAI seem evil if you ask me. Just another company exploiting people who are already being exploited.
In that context, there is no way for OpenAI not to be evil, since they are (by definition) only one option in the market. In fact, taking your argument to the extreme, there is no way to offer jobs in Kenya as the first company to offer jobs would either be exploitative by paying minimum wage or exploitative by being the only real option. Going from that, paying a higher wage just worsens the situation, as it makes the alternatives even less feasible.
That being said, I do get where you are coming from. But it is not a good point to accuse OpenAI on, as they are making the situation better by offering options at a (for a Kenyan level seemingly reasonable) rate and they really don't have any other option[0].
[0] Except maybe paying Americans a lot of money for the job, but I find it morally hard to argue that they should pay US citizens a lot of money instead of paying Kenyans (comparatively) good money, even leaving aside economical feasibility.
Probably the most important measure a company could take to prevent lasting harm with this kind of work would be to spread it around a whole lot more than just 36 people. The real risk of long term impact here would probably be with persistent exposure to it all day long. Most people can handle reading or seeing some graphic stuff with proper mental preparation for it but to see nothing but that day and day out would quickly wear you down.
>Is it ever appropriate or ethical to ask humans to voluntarily subject themselves to traumatic experiences in exchange for compensation?
Probably not. Yet people still voluntarily sign up for military service around the world by the millions, and they do so for a bunch of personal, family, idealistic, cultural, and societal reasons that are hard to reduce to a few easy to argue points like a lot of people online try to do with stuff like this.
Personally, I think it's admirable to hold the ideal that we should like to never offer jobs like this, we should also like to never offer jobs that involve going to war, cleaning up hazardous materials, dealing with explosives, working around heavy fast-moving machines, cleaning sewers, or a myriad of other terrible experiences either; but we're probably not in a position to make those better choices just yet. Until then, people are willing to do these things for a dozen different reasons per person, only two of which are the pay and support they get from the employer.
The difficult part, where you'd get the most out of human labeling of data, is the grey area where the models diverge or are uncertain.
"That month, Sama began pilot work for a separate project for OpenAI: collecting sexual and violent images—some of them illegal under U.S. law—to deliver to OpenAI"
"Sama delivered OpenAI a sample batch of 1,400 images. Some of those images were categorized as “C4”—OpenAI’s internal label denoting child sexual abuse—according to the document. Also included in the batch were “C3” images (including bestiality, rape, and sexual slavery,) and “V3” images depicting graphic detail of death, violence or serious physical injury, according to the billing document."
In this case is text. There’s no graphic depictions of anything. There could be foul, abusive, or racist language. But that’s much less difficult to deal with.
Its not exploitation if you're paying people higher than their local cost of living and higher than other local jobs. If you're just appalled about people being paid an amount of money that to YOU, based on your cost of living isn't fair because you'd want to make more, then be appalled at capitalism as a whole and how much work is being hired outside of company's originating countries. The production of almost everything globally is outsourced to locales where the cost to compete is lower than in the originating country.