Hacker News new | ask | show | jobs
by jsimzeroone 1242 days ago
Neal Stephenson's "The Diamond Age" has a fake AI "character" -- a book that appears to be AI but is really operated by low-paid humans in 3rd world countries (reminiscent of the Mechanical Turk, a supposed chess playing automaton that actually contained a small person).

There's an old observation from Arthur C. Clarke, that sufficiently advanced technology seems like magic. One thing that learning how magic tricks are performed taught me is that magicians typically do their fake magic by doing an unreasonable amount of work behind the scenes -- "magic" in the real world is often just doing a large amount of work that people don't realize is happening.

Given all that it seems appropriate that the new "real world magic" -- ML systems imitating intelligence -- really rest on a lot of hidden work by human beings. Just like magical devices like iPhones exist due to a lot of surprisingly cheap labor. Imagining otherwise is like imagining that the delicious food from a 3-star kitchen just appears from the chef's mind, without the help of all of the low-paid kitchen workers, farm workers, etc. that in reality do most of the work.

10 comments

> a book that appears to be AI but is really operated by low-paid humans in 3rd world countries

Uhm actually :) the AI definietly writes the text itself, and takes care of Nell, and senses the environment around itself, but for plot reasons it can’t do voice synthesis. So it employs humans to read up the words. At least until Nell learns to read.

So it is not just appears, but it is in fact an AI, with a veneer of human voice on it.

In Diamond Age the book is being performed by a skilled actor who is voicing many characters using a script entirely generated by the AI.
You perfectly illustrate the problem with typical western thinking. The workers you refer to may be low paid when compared to the US and other western countries, but they are highly paid in their respective countries. A lot of the time, the jobs (ie: iPhone assembly, AI related) are highly sought after because they are a great alternative to the other jobs that the workers with their skill set can get. It is also a great way to get a step up the job ladder for them and new acquire skills.
> A lot of the time, the jobs (ie: iPhone assembly, AI related) are highly sought after

This description alsp applies to mining toxic substances by hand and high-end prostitution.

The artisan cobalt-mining by hand is relatively popular, and kills you withing 10 years. It is not difficult to explot desperate people. I dont think we should be whitewashing it

The woman who started Sama (mentioned in the article) explicitly started the company to help people in those countries. Her entire life appears to have been directed toward helping people in Africa, she had a history of it. She wasn't there to exploit people.
You mean the same Sama that charged OpenAI $12.50 per hour for a contract and paid their African contractors $2 or less an hour?

https://time.com/6247678/openai-chatgpt-kenya-workers/

Discussed 2 days ago:

https://news.ycombinator.com/item?id=34426421 (570 comments)

(Warning: Shitshow)

Time... aren't exactly being truthful. They constantly refer to "take-home", rather than pre-tax. The all in cost for a person includes pre-tax wages, often taxes on top of that, desks, computers, managers, electricity, office space, benefits and so on. Common sense suggests they are likely making 30% or so margins. If it were much more than that a competitor would eat their lunch. Look at the publicly listed services companies like DXC, Wipro, Accenture - margins are 20-30%.
But it is exploitation though, right? If a bunch of people in a western company say "well, we could just hire people in africa because that would significantly reduce our costs," isn't that an exploitation of cheap labor in africa?
I guess it depends on your definition of "exploit".

If OpenAI had to pay more, they would have gone with another option. It's challenging to work across time zones, across cultures, across language barriers. Working with folks in Reno, NV or somewhere in the southern states of the US would have been the choice for OpenAI at a much higher price.

It's a competitive world. On the surface the Sama founder knew that and realized the options were higher wages for these folks in Africa, or none. The choice of even higher wasn't actually on the table.

> If OpenAI had to pay more, they would have gone with another option... Its a competitive world.

In 2000~s rating agencies rated subprime mortgages as AAA-bonds, causing the global financial crisis. If they rated the bonds as junk, the banks would go to another rating agency. Its a competitive world.

Therefore defrauding all of us was the right thing to do?

Just because the problem is inherent in the system and you individually can't change it, does not mean you cannot acknowledge the system's fault's.

Is it exploitation to buy things from poor people?
Is it exploitation to buy a kidney from a homeless man for $500?
It can be, if you're taking advantage of them being poor in order to pay them way less than you would pay others.
It would be much worse if they paid way above the local wages. It would trigger corruption. You'd could up with a black market in applications for the jobs, or protection racket, or highly paid leaks of the job interview problem sets, imagination has no limit.

You can say it's OpenAI's duty to make sure all these things don't happen. But they are not there to police the local society. They are there to run their own business. They don't have the competency to make sure corruption does not happen.

No, it's use of cheap labor in Africa.
All business is about exploiting someone somewhere for your benefit, no exceptions. The only question is whether that exploitation is within tolerable limits.
> All business is about exploiting someone somewhere for your benefit, no exceptions.

Where did you learn that? I don't think it's at all true. It seems maybe you have a no-true-scotsman definition of 'exploited', so that no evidence against your claim would change your mind.

Picture a baker, who makes bread for people, who get bread, the baker gets money. Where is the necessary exploitation? I can't imagine where your confidence - "no exceptions" comes from. There are no win-win exchanges in the world, and none even possible? I'm not a huge fan of capitalism but that seems absurd.

I think your example explains why the article is a potentially gray area of exploitation.

I think it's clear exploitation when you offer someone employment that adversely impacts their rights. So artisan cobalt-mining is exploitative, because of it's health effects, would go against the UN's definition of "the right to work in just and favourable conditions"[1]. However, I'm not sure what, if any, rights are being compromised by the topic in the article. Maybe there's a case that it's unjust because of the asymmetry in the value created and payment. If there are some rights abuses, then it becomes a clearly exploitative endeavor.

[1] https://www.un.org/en/global-issues/human-rights

Living in Asia, I agree with the OP and find your total switch of topics in order to gain status points on HN gross. People in Asia desire good factory jobs manufacturing iPhone and what not. What the hell is wrong with that? Why you switch this to mining cobalt by hand is totally dumb. Also where are your sources? Mining cobalt by hand? Any mining operation in Asia will use the latest tools available. You think people here are running around in huts, scavenging cobalt and other precious metals from computer motherboards discarded in the year 1998? Get with it!
Exploitation involves, by definition, giving the exploited something they need. If you don't have food, and someone gives you food in return for slave labor, they are exploiting you.
If people in a poor country are offered jobs in a factory, and they prefer those jobs over subsistence farming (since the conditions are better and pay is better), who exactly is worse off as a result? Sounds like a profitable trade to me given both parties walk away happy.

Outsourced factory jobs are the mechanism by which previously poor countries like Taiwan and China (in many ways) have been pulled out of poverty. The process is happening before our eyes in Vietnam right now.

Do you think you’re doing poor people a favour by denying them well paid jobs? Should we do the same in the west, and have companies fire all our poorest employees?

I can hardly think of a more cruel policy.

> If people in a poor country are offered jobs in a factory, and they prefer those jobs over subsistence farming (since the conditions are better and pay is better), who exactly is worse off as a result?

You aren't understanding the meaning of "exploitation". See the GP.

> both parties walk away happy

Are they actually happy? Or are they exploited? That's the issue.

>> If people in a poor country are offered jobs in a factory, and they prefer those jobs over subsistence farming (since the conditions are better and pay is better), who exactly is worse off as a result?

> You aren't understanding the meaning of "exploitation". See the GP.

My question stands regardless of the meaning of the word "exploitation". I didn't use that word in my question. - Which I note you dodged answering. Who is worse off? Can you point to them?

>> both parties walk away happy

> Are they actually happy? Or are they exploited? That's the issue.

Why are they mutually exclusive? If being exploited isn't something with negative connotations, why can't I be "exploited" and be happy at the same time? If I hire a plumber to fix my leaky sink, who cares what the definition of the word exploited is? He's happy for the work, and I'm happy for my sink being fixed. We all walk away better off.

Was he exploited? Was I? Who cares when its consensual and we're both better off?

This argument about what "exploit" means seems pointless and unrelated to the actual topic - which seems to be, is it moral to hire someone in a poor country to do work for you. My claim is that not sending work to poor countries is often even more cruel - because it keeps people in poverty.

> This argument about what "exploit" means seems pointless and unrelated to the actual topic - which seems to be, is it moral to hire someone in a poor country to do work for you. My claim is that not sending work to poor countries is often even more cruel - because it keeps people in poverty.

Exploitation is not a pointless issue; it harms many millions, maybe billions of people.

The options are not only A) Exploit people, or B) Don't hire anyone. We can and should hire them in ways that aren't exploitative.

This is another common problem with tropical western thinking -- forgetting the historical context. Colonialism happened, and the people are so poor and desperate now because of things the west did in the past.
This does explain a lot of the difference in development, especially in very poor countries. And this isn’t just a result of “old”, traditional colonialism, but imperialism in the form of meddling in their internal politics. Guatemala is a great example, having had its democracy destroyed by Allen Dulles and his CIA in the 1950s. But you can also take a more recent example like the 2009 coup in Honduras, backed by Hillary Clinton.
As someone who does not live in the USA Iam very glad H Clinton did not become president

Impossible to know but I expect she would not have stayed out of the Syrian war.

It was she who was largely responsible for the debacle in Lybia

So Karen on using all that power and such a short term thinker

Trump was a catastrophe for you in the USA but not really for us.

Please elect an isolationist....

The DC establishment will do anything to prevent an isolationist from taking power-- that would begin the rollback of the sprawling empire from which they derive their largess. Trump was terrible domestically, and he was only isolationist in some areas (certainly less damaging overall than the Karen). He also appointed some usual suspects from the Deep State who are extremely interventionist (John Bolton, ffs).
The west doesn’t have a monopoly on evil. Plenty of Africans participated in the slave trade, and sold their brothers and sisters. And still do today.

There's something weirdly racist about the idea that only people in the west (or American diplomats) have any agency in the world. That America somehow is so powerful that y’all created the modern world. That local people and local leaders have no capacity for determining their own future.

That corruption in the police in South Africa is America’s fault. That the problem of gangs in Mexico was created by the US and that the Mexican gang members are blameless when they decapitate people. That Americans were the people who voted in Bolsenaro. Or the west are responsible for Japanese aggression in WW2.

Maybe ending colonial ideas starts at home.

> ...magicians typically do their fake magic by doing an unreasonable amount of work behind the scenes...

I probably heard this in a Penn & Teller interview, but the idea is much older... One of the keys to making a good illusion is to put far more work into it than people would think is reasonable, so it won't occur to them to think along the lines of what you actually did.

> Mechanical Turk

And also inspiration for Amazon's Mechanical Turk. [0]

[0]: https://www.mturk.com/

Most of smartphones' magic exists due to very expensive labor of hardware and software engineers. Production of parts and assembly is 1-2% of the total work done.
Reminds me of that "virtual youtubers" are actually humans behind the skin
>reminiscent of the Mechanical Turk, a supposed chess playing automaton that actually contained a small person

Reminiscent of the Amazon Mechanical Turk, a current real-life example of exactly they.

> a book that appears to be AI but is really operated by low-paid humans in 3rd world countries

I've seen this in real life before, no need for a work of fiction.

Diamond Age was first published in 1995.

I don’t think I am out on an intellectual limb believing very very few people would have seen an AI written book back then.

Not as a rule though - so many ML systems are utilizing data that is streaming in from passive sensors or transactional streams and is not human curated at all. The human aspect isn’t an intrinsic property of ML or even these algorithms only particular applications (and I would guess a minority of applications too).

Given that, it seems to be a clear miss to apply that logic generally. I have to believe it most likely stems from a lack of basic understanding and competence on the authors part.

Edit: Here are two examples that I have personally worked on:

Global calculations of weather composite reflectivity using 20 years of historical satellite imaging data from NASA

Superior RF demodulation (under certain circumstances) using RF transmissions as received and sent.

Both of these utilize modern ML imaging models, neither require any human labeling only streaming data (which in these cases began collection long before modern ML techniques were in widespread use). The applications in the natural sciences are endless not to mention the applications more on the business intelligence side using transactional data. Only in specific cases is human labeling required but because of the high cost of that data it is by its nature dwarfed by that which is collected naturally (not to mention often error prone). It is for that reason that techniques to ingest data that is more and more natural to collect are growing in favor.

>> I have to believe it most likely stems from a lack of basic understanding and competence on the authors part.

That is unlikely, given that one of the authors is Timint Gebru. I'm quoting below select passages from her wikipedia page indicating her background:

In 2001, Gebru was accepted at Stanford University.[2][5] There she earned her Bachelor of Science and Master of Science degrees in electrical engineering[8] and her PhD in computer vision[9] in 2017.[10] Gebru was advised during her PhD program by Fei-Fei Li.[10]

Gebru presented her doctoral research at the 2017 LDV Capital Vision Summit competition, where computer vision scientists present their work to members of industry and venture capitalists. Gebru won the competition, starting a series of collaborations with other entrepreneurs and investors.[11][12]

Gebru joined Apple as an intern while at Stanford, working in their hardware division making circuitry for audio components, and was offered a full-time position the following year. Of her work as an audio engineer, her manager told Wired she was "fearless," and well-liked by her colleagues

https://en.wikipedia.org/wiki/Timnit_Gebru

I didn’t say they were uneducated but an audio hardware engineer does not imply a good working knowledge of industry trends in ML applications.

Regardless, my point still stands, they completely ignore (willingly or ignorantly) that human labeled data is not intrinsic to ML or even the algorithms themselves and in all likelihood is a small minority of datasets used by modern ML applications. To then apply that critique generally to ML shows ignorance and a misunderstanding of the ecosystem.

Gebru is not an audio hardware engineer. I call your attention to this passage I quoted above:

Gebru presented her doctoral research at the 2017 LDV Capital Vision Summit competition, where computer vision scientists present their work to members of industry and venture capitalists. Gebru won the competition, starting a series of collaborations with other entrepreneurs and investors.[11][12]

And to the fact that she got her PhD in computer vision, i.e. the main area of AI research that the article seems to be criticising.

Her work experience is as an audio engineer - but again it doesn’t matter what her credentials are, she is wrong regardless and you are ignoring my whole point. She shows her ignorance of the subject matter (again willingly or not) when she applies her critique generally at Ml and not just at these specific applications - not sure how many times I need to say that.
>> Her work experience is as an audio engineer

Her PhD research is in computer vision and she and her co-authors are writing mainly about computer vision, but you spoke of "a lack of basic understanding and competence on the authors part". That is clearly incorrect and I don't understand what saying the same thing many times will change about that.