Why Big Tech Pays Poor Kenyans To Teach Self-Driving Cars (bbc.com) 78
Each day, thousands of people from places like Kibera, Africa's largest slum and one of the toughest neighborhoods on earth, commute to an office of Samasource in the east side of Nairobi. The San Francisco-headquartered company occupies four floors of a business park building, with vast banks of computers being used for the job of training data. Google, Microsoft, Salesforce and Yahoo are among the clients of Samasource. What exactly do these people do at Samasource? Its clients won't say, but BBC reports that the "information prepared here forms a crucial part of some of Silicon Valley's biggest and most famous efforts in AI." From the report: [...] Brenda loads up an image, and then uses the mouse to trace around just about everything. People, cars, road signs, lane markings -- even the sky, specifying whether it's cloudy or bright. Ingesting millions of these images into an artificial intelligence system means a self-driving car, to use one example, can begin to "recognise" those objects in the real world. The more data, the supposedly smarter the machine. She and her colleagues sit close -- often too close -- to their monitors, zooming in on the images to make sure not a single pixel is tagged incorrectly. Their work will be checked by a superior, who will send it back if it's not up to scratch. For the fastest, most accurate trainers, the honor of having your name up on one of the many TV screens around the office. And the most popular perk of all: shopping vouchers.
It's the kind of technological progress that will likely never be felt in a place like Kibera. As Africa's largest slum, it has more pressing problems to solve, such as a lack of reliable clean water, and a well-known sanitation crisis. But that's not to say artificial intelligence can't have a positive impact here. We drove to one of Kibera's few permanent buildings, found near a railway line that, on this rainy day, looked thoroughly decommissioned by mud, but has apparently been in regular use since its colonial inception.
Almost exactly a year ago, this building was the dividing line between stone-throwing rioters and the military. Today, it's a thriving hub of activity: a media school and studio, something of a cafeteria, and on the first floor, a room full of PCs. Here, Gideon Ngeno teaches around 25 students the basics of using a personal computer. What's curious about this process is that digital literacy is high, even in Kibera, where smartphones are common and every other shop is selling chargers and accessories, which people buy using the mobile money system MPesa.
It's the kind of technological progress that will likely never be felt in a place like Kibera. As Africa's largest slum, it has more pressing problems to solve, such as a lack of reliable clean water, and a well-known sanitation crisis. But that's not to say artificial intelligence can't have a positive impact here. We drove to one of Kibera's few permanent buildings, found near a railway line that, on this rainy day, looked thoroughly decommissioned by mud, but has apparently been in regular use since its colonial inception.
Almost exactly a year ago, this building was the dividing line between stone-throwing rioters and the military. Today, it's a thriving hub of activity: a media school and studio, something of a cafeteria, and on the first floor, a room full of PCs. Here, Gideon Ngeno teaches around 25 students the basics of using a personal computer. What's curious about this process is that digital literacy is high, even in Kibera, where smartphones are common and every other shop is selling chargers and accessories, which people buy using the mobile money system MPesa.
Good idea. (Score:2)
Re: (Score:1)
From what I can tell, Kenyans as a group are not good drivers.
Re: (Score:3, Funny)
Alexa: "It's right next to your tax returns."
Re:Couldn't resist (Score:5, Informative)
Re: (Score:1)
It wasn't his book; it was on some university pamphlet that Obama himself did not write. An old fashioned typo; not a conspiracy.
Re: (Score:2)
Keynesians save during boom times.
There are no Keynesians, only spenders with an excuse.
Debt & Savings [Re:Couldn't resist] (Score:2)
GOP didn't used to be anti-Keynesian; they were ambiguous on the topic. Their recent criticism of it seems to be political in nature, not inherent doctrine*.
To me, Keynesian economics is basic common sense: save up during the good times so you have some spending room during bad times, such as for stimuluses. Everyone should do that with their money (unless you have a fatal disease). I welcome logical arguments against it.
* A few libertarian-leaning Republicans will say that ALL stimuluses are bad because th
Re: (Score:2)
If I ever encounter an actual Keynesian, I expect it would be in some university setting.
All I see are spenders with an excuse.
Re: (Score:1)
The real problem is that a politician is rewarded far more for short-term benefits than long-term benefits. Thus, there's an incentive to spend on goodies (services & tax cuts) now and let later generations of politicians deal with the fallout. It also has happened with pensions, not just gov't debt.
We need a different way to ensure fiscal discipline.
Because (Score:3)
Just my 2 cents
let me guess (Score:1)
guinea pigs?
Why? Money. (Score:5, Insightful)
Why does big tech employ third-world workers to do repetitive menial tasks? Because they're cheap.
I don't think it's a bad thing. I've lived in some of the nicer areas in Africa, where $0.10 (US) buys a full meal at a restaurant. If a tech company can establish an office, and dump a few tens of thousands of dollars into their economy, those workers will be some of the wealthiest in the area.
It's a paying job, fairly stable, and less likely to kill than many other jobs in the area.
Re: (Score:2)
Should we? Well, sure.
Can we? Nope.
For most of rural Africa, the math and logic they need to get through life is limited to making trades at the market and keeping their farm running. To make a VERY broad generalization, there is little appreciation for the skills of a secondary-school education, let alone a collegiate-level software development skill set. I've seen parents yell at teachers for "wasting time" with reading.
This new work might lead to an appreciation of "white-collar" skills, diversifying the
Re: (Score:3)
For most of rural Africa, the math and logic they need to get through life is limited to making trades at the market and keeping their farm running.
The "math and logic" to keep a farm running (translation: operate a small business) is not trivial.
Re: (Score:2)
True, but it's also not formalized or rigorously defined to the degree necessary as a foundation for software development. Instead of measuring and defining risk factors and consumption rates, it's looking at a field and saying "that should be enough".
It's also worth noting that the culture is, in some ways, a lot more forgiving than Western standards. In the West, if you screw up your financial estimates, you fail, go bankrupt, sell the farm, and try to find a different job. In rural Africa, if you don't h
Re: (Score:2)
It raises average salaries and increases community wealth. That will result in some people lording it over their neighbours but will also result in people investing in improving their community.
Taking people out of subsistence living is a basic first step towards a large range of improvements, including education and gaining a more complex range of skills.
Re: (Score:1)
Re: (Score:2)
Also, it tends to improve the infrastructure the business depends on like electricity and network connections and it makes getting enough of an education to get these jobs make sense. A lot of the time the poorest areas are trapped in a catch-22, they can't get jobs and without jobs there's no money to improve anything. Obviously companies are doing this for profit and not charity but on the tail end of that exploitation it's actually more like an opportunity.
Re: (Score:2)
Trickle-down economics works for infrastructure, not for money.
Re: (Score:2)
I did smirk a little bit at an article about this where the (white, affluent, DISTANT) manager said that she had to be careful not too pay them too much to wreck the local economy.
I mean, I get it, on a superficial level she's right, but they're paying them $5 a day. I'm going to guess her bonus on not-paying-them-much has little to do with preserving the local economics.
Re:Why? Money. (Score:4, Informative)
That is actually a thing.
For comparison, where I was, an average monthly wage was (at the time, in a rural area) 40 USD*. For a single company to pay an American-standard wage there would be like a company opening an American office paying an average of $7.5 million annual salary today. It sounds great, but knowing that a major employer in an area pays such high rates opens the door for local hyperinflation, because everyone who doesn't work for that company knows that the folks who do will be able to pay higher prices. Along with that inflation comes an influx of scams and crime, because the reward is worth far more than the expense of running their scam.
The most I've seen in person was a mining company that paid around $150/month on the same rural region. That's about equivalent to a $300,000 salary in America today. The results were about what's seen in San Francisco... Spiking real estate prices, heavy commuter traffic, and intense pressure to sell property and move out of the local area to cash in on the bubble.
From what I found with a quick search, the average monthly salary in Kenya is $76, which works out to about $3.50 per day. Paying $5 per day (roughly $105/month) doesn't seem that bad, superficially or not.
* All monetary amounts mentioned are US equivalent, regardless of local currency, and not adjusted for inflation, locale, industry, or much of anything. I'm also not completely sure of my math, but the general idea should be correct.
Re: (Score:3)
No kidding. The invisible hand.... a business wants very cheap labor, so it ends up giving money to very poor people.
The industrial revolution was partly about people having horrible jobs in factories instead of starving in a hut. But for the next factory to attract (better) workers, it has to pay a bit more..... and so on until even poor people have TVs and $100 dollar sneakers.
The Future (Score:2)
Re:The Future (Score:5, Insightful)
Do these Kenyans know that in 15 years their cab and truck driving jobs are going bye-bye?
That will happen whether or not they work for companies classifying images.
And if you're poor in an impoverished city in Africa, you're going to care more about getting a job, any job, now, and not what is going to happen to possible future jobs fifteen years down the road.
The article claims they are not (Score:2)
Do these Kenyans know that in 15 years their cab and truck driving jobs are going bye-bye?
Looks like you didn't read the summary closely enough:
It's the kind of technological progress that will likely never be felt in a place like Kibera
Sounds like the article writer thought it would never take away Kenyan jobs.
Personally I think it far more likely it will reach Kenya, as cars that do not need drivers can be way more widely dispersed than cars that require humans, and can operate more cheaply. Also maybe t
Re: (Score:2)
Sounds like the article writer thought it would never take away Kenyan jobs.
No, sounds like the writer is pushing their idiotic prejudices. After all..
smartphones are common and every other shop is selling chargers and accessories
Oh look, people are already accessing modern electronics.
Re: (Score:2)
Do these Kenyans know that in 15 years their cab and truck driving jobs are going bye-bye?
You are implying that automation is bad for an economy.
If you open your eyes and look at the world around you, you will see that this is the exact opposite of reality.
Re: (Score:2)
They don't commit suicide when their mundane task is automated, they jump when their mundane task isn't automated and they have to work 12 hour days at unpleasant tasks.
Re: (Score:3, Informative)
So far.
But automation will eventually displace all work. Once you do that, there's no star trek-esque space exploration. It's all endgame capitalists with all the money in the world, and the rest with nothing.
What we call "menial jobs" is what MOST PEOPLE in the world do, and are fine doing. The "educated, creative types" are a very rare exception. We might stand a chance in a fully automated world. But "most people" won't. They just don't have the mental ability to.
Automation doesn't "create more jobs". Au
Re:The Future (Score:5, Insightful)
But automation will eventually displace all work.
That will require human level AI which is pure sci-fi. It may happen someday, but if it does, the changes to our existence will be so profound and unpredictable, that "jobs" will likely be the least of our concerns.
It's all endgame capitalists with all the money in the world, and the rest with nothing.
It was once predicted that only "the rich" would be able to afford cars. The same was predicted about computers, washing machines, dishwashers, etc.
Now you are predicting the same about robots, replicators, blah blah blah. Whatever.
There is no reason that mass produced robots should be unaffordable to the masses. Once the design cost is sunk, it will be cheap stamped or molded parts, and software (marginal cost: $0). Anyone that can afford a refrigerator today, will be able to afford a household robot-maid and fabricator a decade from now.
Automation doesn't "create more jobs". Automation destroys jobs
I see. So that explains why America, Europe, and East Asia are starving, while countries that wisely avoided automation, like Ethiopia, Niger, and Afghanistan are so prosperous. Whatever.
Re: (Score:2)
The industrial revolution happened because it allowed to lower costs, and this in turn made it so more people could afford things. Since now more people bought more things, there was a demand for different jobs. Now you automate the new jobs. This works only for so long. There is a point where you will have automated all work, and once this happens, people lose their job. This is a downward spiral. You don't need "human level AI", you just need to displace a big enough percentage of works to tip the system
Re: (Score:3)
This works only for so long.
There are billions of people living on less than $5 per day. The time when "people have all the stuff they need" is a long, long way off.
Even when people have all the stuff they want, there are still services, which provide 80% of jobs in developed economies. Nobody wants to eat in a fancy restaurant just to be served by a robot. If they want automated food, they would order take-out.
you just need to displace a big enough percentage of works to tip the system off.
That already happened when agriculture was mechanized. Instead of starving, living standards soared, and today we are in
Re: (Score:2)
More people eat at McDonald's than they eat at all fancy restaurants in the world, combined.
That's the thing. There are only a few "elite" types that will be able to "afford nice things", and too many people willing to serve them. That's such a huge oversupply that will make salaries almost zero.
Re: (Score:2)
There's no doubt that automation makes things cheaper but only if you stay employed at a constant wage. If you look at the US the median [cleveland.com] wage, like what a typical employee will make that has been flat for 50 years. Now tons of jobs have been automated since the 60s, why haven't the wages skyrocketed? Because it's supply and demand, your typical worker is barely able to keep the purchasing power he has. You see the top 5% and 10% are pulling ahead, never mind the top 0.1%-1% but those that aren't on the grav
Re: (Score:2)
You are talking about inequality, not unemployment.
Does automation cause inequality? Yes, there is plenty of evidence for that.
Does automation cause unemployment? No, it causes some short-term dislocation, but not long term unemployment.
There is no evidence that automation "destroys jobs".
Even for inequality, there is little evidence that there are "losers". The top 20% gained a lot, the middle 60% gained some, and the bottom 20% were stagnant, but still no worse off than before.
The rise in inequality ha
Re: (Score:2)
Automation doesn't destroy jobs?
Did you know before the invention of the alarm clocks we had people knocking on your window to wake you up?
Did you know before the invention of electric light, you had people turning on and off all street lights?
Plenty of jobs have been destroyed by automation.
Your point on "short-term dislocation" is the key here. So far we've increased the demand fast enough to reconvert the lost jobs. But at some point we won't be able to. This is what I'm trying to make you understand. Yo
Re: (Score:2)
Oh we are heading for collapse in the long term. The rich will NOT give us "just enough". The rich won't give us anything. Rich people build higher walls to keep the poor outside.
This is like any other failed communist state (like Venezuela right now), where you have a top 1% government ultra-rich, a 10% "well fed" military to keep the poor at bay, and the rest of the population slowly starving (venezuelans have lost several kg weight in average in the last decade). This is what we're heading for. But the u
Re: (Score:2)
Honestly, their cab and truck driving jobs are just fine for quite a bit longer. There's a substantial period where a $2/hour human > an expensive AI > a $20/hour human. There are millions upon millions of manual jobs in China that *could* be automated, but not for cheaper than subsistence workers.
Re: (Score:2)
So (Score:1)
Does this mean self driving cars will drive like poor Kenyans?
If AI can work in Kenya (Score:2)
Re: (Score:2, Funny)
Alabama: Challenge accepted.
good thing that criminal liability can't be EULA (Score:2)
good thing that criminal liability can't be waved with an EULA or that logs / other data can't be locked in an NDA an a criminal trial. But it may take something really bad like an auto drive truck wiping out a school bus full kids to get it to a court room.
Great to see (Score:3)
Re: (Score:3)
also lets them Outsourc liability (Score:2)
also lets them Outsource liability???
What happens when the dial down the match % and you have an miss match that leads to an death?
I think I get it... (Score:2)
Lowest common denominator?
Samasource is Nonprofit (Score:5, Informative)
Missing from the summary is the fact that Samasource is a non-profit focused specifically on providing opportunities for some of the worlds poorest people.
wiki [wikipedia.org]
Image-ine that! (Score:2)
I have to wonder, are these AIs really learning to generalize? Or are they having millions of examples, so many that any live image must be within a tiny delta of a known image?
Re: (Score:2)
I have to wonder, are these AIs really learning to generalize? Or are they having millions of examples, so many that any live image must be within a tiny delta of a known image?
This is a well known problem, called "overfitting". You can test it by only training your model on a subset of the images, and then see how it does on the remaining ones. There are various techniques to combat overfitting.
Because ... (Score:2)