How machines can learn gender bias

Several readers wrote to me saying that they found my previous post on algorithms propagating social inequality interesting. So here is a link on how machines (algorithms) can pick up gender bias in literature.

“This machine read 3.5 million books then told us what it thought about men and women” (from World Economic Forum):  VideoArticle

 

Tech neutrality (or not) of Facebook

A while ago, I wrote a post about the myth of tech neutrality.  Here I’m sharing link to a long article from Wired about Facebook and it’s own slow realization of the same:

It appears that Facebook did not, however, carefully think through the implications of becoming the dominant force in the news industry. Everyone in management cared about quality and accuracy, and they had set up rules, for example, to eliminate pornography and protect copyright. But Facebook hired few journalists and spent little time discussing the big questions that bedevil the media industry. What is fair? What is a fact? How do you signal the difference between news, analysis, satire, and opinion? Facebook has long seemed to think it has immunity from those debates because it is just a technology company—one that has built a “platform for all ideas.”

This notion that Facebook is an open, neutral platform is almost like a religious tenet inside the company. When new recruits come in, they are treated to an orientation lecture by Chris Cox, the company’s chief product officer, who tells them Facebook is an entirely new communications platform for the 21st century, as the telephone was for the 20th.

Over time…

…WHILE FACEBOOK GRAPPLED internally with what it was becoming—a company that dominated media but didn’t want to be a media company—Donald Trump’s presidential campaign staff faced no such confusion. To them Facebook’s use was obvious. Twitter was a tool for communicating directly with supporters and yelling at the media. Facebook was the way to run the most effective direct-­marketing political operation in history.

The supposedly neutral “platform” was actually enabling certain kinds of use:

… [security researcher Renee] DiResta  published an article comparing purveyors of disinformation on social media to manipulative high-frequency traders in financial markets. “Social networks enable malicious actors to operate at platform scale, because they were designed for fast information flows and virality,” she wrote. Bots and sock puppets could cheaply “create the illusion of a mass groundswell of grassroots activity,” in much the same way that early, now-illegal trading algorithms could spoof demand for a stock. …

THE way the Russians used the platform was neither a surprise nor an anomaly. “They find 100 or 1,000 people who are angry and afraid and then use Facebook’s tools to advertise to get people into groups,” he says. “That’s exactly how Facebook was designed to be used.”

And even as the world realized Facebook (and fake news’) role in 2016 US Presidential election, the company was slow to come to terms with it:

…For the first time, insiders really began to question whether they had too much power. One employee told WIRED that, watching Zuckerberg, he was reminded of Lennie in Of Mice and Men, the farm-worker with no understanding of his own strength.

The supposedly neutral “platform” was actually enabling certain kinds of use:

… [security researcher Renee] DiResta  published an article comparing purveyors of disinformation on social media to manipulative high-frequency traders in financial markets. “Social networks enable malicious actors to operate at platform scale, because they were designed for fast information flows and virality,” she wrote. Bots and sock puppets could cheaply “create the illusion of a mass groundswell of grassroots activity,” in much the same way that early, now-illegal trading algorithms could spoof demand for a stock. …

THE way the Russians used the platform was neither a surprise nor an anomaly. “They find 100 or 1,000 people who are angry and afraid and then use Facebook’s tools to advertise to get people into groups,” he says. “That’s exactly how Facebook was designed to be used.”

The way technology is designed, the way its algorithms function, impact what gets ‘rewarded’ and grown:

“If we just reward content based on raw clicks and engagement, we might actually see content that is increasingly sensationalist, clickbaity, polarizing, and divisive,” she says. A social network that rewards only clicks, not subscriptions, is like a dating service that encourages one-night stands but not marriages.

And:

The past year has also altered Facebook’s fundamental understanding about whether it’s a publisher or a platform. The company has always answered that question defiantly—platform, platform, platform—for regulatory, financial, and maybe even emotional reasons. But now, gradually, Facebook has evolved. Of course it’s a platform, and always will be. But the company also realizes now that it bears some of the responsibilities that a publisher does: for the care of its readers, and for the care of the truth. You can’t make the world more open and connected if you’re breaking it apart. So what is it: publisher or platform? Facebook seems to have finally recognized that it is quite clearly both.

 

The myth of tech neutrality

In my class, I sometimes like to show students a disposable plastic water bottle, one that is so thin that just holding the empty bottle dents it.  I ask them whether a plastic water bottle represents technology. Slowly but inevitably we conclude that it is – perhaps not a particularly advanced one by modern standards, but certainly one that changed how humans transport and consume water and other liquids.

Next, I show them another water bottle: One made of steel. I ask them, whether technology is neutral?  Do these water bottles assume different lifestyles?  One or two students will point out the obvious – that one is intended for reuse while the other isn’t.

That is usually a good entry point for a discussion on the assumptions underlying our use of technology – each product assumes certain lifestyles and what software designers call “use cases”.  A thin plastic water bottle assumes a single use scenario while a steel one imagines that the user will carry such a bottle with her everywhere she goes and that there will be clean water available to refill as needed.

I ask them the same question about apps, cars, and public transport – are these technologies neutral?  It takes a while for students to realize that in each of these cases a certain type of user has been imagined.  Even after 100 years of car design & manufacturing experience, car-designers haven’t made cars which meet women’s need to put their handbags somewhere within reach.  Thousands of years of clothing design has not made designers realize that women need pockets as much as men do.  Apple’s health app, when first released, did not include the one thing every adult woman tracks – her menstrual cycle.  And that’s just a shortlist of issues from a gender perspective.  One could add even more problematic issues relating to usability of products for those with physical disabilities.

These “technologies” which we take for granted aren’t neutral; they are manifestations of the imaginations of their designers – in terms of who will use them, what their lifestyle is (or will/ should be).  The designers of a kitchen assume a “standard” height for cooks.  The designer of a mobile phone assumes that the user is literate.  And the designer of a nuclear bomb assumes that there will be a situation during war where one side will feel that it is justified in killing millions of innocent citizens simply because they are living on the other side of a political boundary.

Recently I listened to a Radiolab podcast on AI and ethics.  They started with the usual trolley problem:  An out-of-control train is hurtling down a track on which 5 workers are working.  The workers will not be able to get out of the way in time.  But you have the option of flipping a lever so that the train is diverted to a different track where one person is standing.  So you have a choice of killing one versus five people.  Most people will select killing one person over five.  To this well-known ethics problem, they added a twist.  Now there is a self-driving car with an AI.  In an unavoidable accident, the AI has to decide whom to kill – one person (its passenger) or five pedestrians.  The situation hasn’t changed, but although in the trolley problem most people want to save 5 people over 1, most buyers of the self-driven car want the car AI to protect them instead of the 5 pedestrians.  Now imagine thousands of self-driven cars making such decisions on behalf of humanity.  Obviously, car manufacturers are going to pay more attention to the needs of the buyer/ passenger and hence the current cars are being designed to save the passenger.

Ask most people if they think technology is neutral and they’ll say ‘yes, it is how one uses it which determines whether it is good or bad’.  Modern technology is so complex that we hardly understand what we are using – whether it is the nuances of facebook feed algorithms, the persistent need to get more ‘likes’ or the ethics principles underlying self-driving cars.  Even non-smart technology like the ubiquitous plastic water bottle surely nudges us towards a different vision of the modern lifestyle than what we might have chosen for ourselves.

When we promote e-learning in school classrooms, we imagine that teachers are unable to teach these subjects effectively, that students can learn equally well from impersonal videos as from human teachers, and that instead of investing in teacher capacity building it is easier, faster and “more efficient” to develop e-content.

Aadhar designers imagine that there are situations where a government needs to identify individuals either for controlling access for certain schemes (e.g. PDS) or tracking actions.

Rather than assuming that such uses of technology are inevitable, we should pause and think about the direction such technologies are nudging us towards.*

Most people, when they say ‘technology is neutral’, what they really mean is that the science behind the technology is universal.  Broadly defined, technology is the application of scientific knowledge for practical purposes.  While the science of making a plastic water bottle may be neutral, deciding that such bottles represent a desirable lifestyle and therefore must be available cheaply is not. It implicitly privileges the privatisation of drinking water supply over the (harder) work of holding governments’ accountable for ensuring that the water coming out of our taps is potable.

I read with great interest the book “Geek Heresy” by my friend Kentaro Toyama, who makes a very convincing argument that technology amplifies but doesn’t solve underlying processes and realities.  For example, e-education amplifies the inequities related to access to education as does e-medicine.

These days it has become fashionable to invest in agri-tech, mobile-based health services, app-enabled government petitions, etc.  Now that we have reached a point where we hardly ever interact with other people without mediation of technology, it is important to occasionally take stock and reconsider: which technological products represent “use cases” which are desirable (piped water in every home?), which ones should be limited to temporary stop-gap measures until longer-term initiatives can take effect (mobile-health apps?) and which ones should be avoided completely (plastic straws?).

Given the pervasiveness of all kinds of technology around us (when was the last time you touched something not made by humans?), these questions should no longer be decided by product companies but as a society. As one commentator on the Radiolab show said in a colourful way, “I don’t want a 20 year old wearing jeans and sipping coke at his computer deciding for all of us.” I have nothing against 20 year olds – in fact, they are the ones with the burden of solving problems that my generation has created.  But the commentator highlights an important point:  These decisions about technology are ones we must make collectively as a society – whether they are about the use of plastic straws, the ethical principles self-driving cars, the algorithms of Google and Facebook feeds, or Aadharification of social security benefits.

* See for example: http://www.wired.co.uk/article/chinese-government-social-credit-score-privacy-invasion

Soil nutrient management

Previously I have written about mobile-phone based personalized alerts for crop management offered by EkGaon Technologies founded by Vijay Pratap Singh Aditya. And about KC Mishra who is developing various software tools for helping farmers manage soil nutrients and select the right seeds. The software is available at his eKutir chain of shops run by micro-entrepreneurs.

And now, we hear that Re:char, a Kenyan company is creating a device called SoilIQ that would fit on an android phone and measure soil nutrients on the spot.

I wonder if these three could be combined to create an end-to-end solution for farmers.

  1. Soil testing using a device like Re:Char’s SoilIQ device
  2. Customized advisory on soil nutrient management through solutions such as eKutir’s tools or EkGaon’s research
  3. Soil nutrient management services via mobile SMS/calls through services such as EkGaon

Of course, there are still lots of issues that make the feasibility of such an end-to-end solution questionable:

SoilIQ measures only moisture content of soil. It is still in design phase. And, the cost of  such a combined service might be too high.  Plus, farmers’ may not be able to afford android devices.  But it is still interesting to imagine such a solution, perhaps leveraging a micro-entrepreneur or NGO as a service provider might solve some of these issues.

Given the deteriorating soil conditions in the country, it would be useful to have such information easily available, complete with advise on corrective measures not just for the macro fertilizers but also micro-nutrients for the soil.

As science begins to prove the link between nutrients in the soil, in the crops grown on these soils and in people who eat these crops, managing soil nutrients becomes an important tool for improving malnutrition in India.

In Search of a “Progressive Farmer”

Santosh excitedly scrolls through many images of his soya fields on his touch screen phone as I watch. The images were taken at various stages of cultivation – I can see images of field preparation with neat rows ready for planting, and photos of young plants after a few weeks of growth.  He scrolls through the images as he talks about the things he has learnt using the internet.

Soya fields in Madhya Pradesh

Most soya farmers in Madhya Pradesh I had met before Santosh, knew that Brazil is the leading producing of Soya in the world.  But Santosh’s knowledge goes deeper. He knows the kinds of micro-nutrient fertilizers used in South America.  Santosh explains that the typical spacing of soya plants in Vidisha is 9 inches but in Argentina it is 18”.  On one portion of his own farm he has tried 12” this year to see if it improves his yields.

A farm inputs shop in Vidisha, Madhya Pradesh

Santosh decided to spread his knowledge through the government’s ATMA program for agri-extension services.

And, he opened an agri-inputs business that sells the same farm inputs that he uses and a few other high quality brands, which he says he selects carefully.

As he talks, Santosh opens a drawer in the shop counter and pulls out a tiny bottle barely three inches high.  This is the Argentinian micro-nutrient he’d discovered through his internet research.  Apparently the micro-nutrient fertilizer helps with root branching of soya plants.  Santosh has tried it on his farm with positive results. And he has sold a lot of it from his shop.

Seeking knowledge

Santosh is an example of the rare entrepreneurial farmers who seek information, experiment on their own fields and take a professional interest in farming.  He is the exact opposite of most farmers I’ve come across (even those with large landholdings) who display the opposite traits. They are unwilling to experiment even with small portions of their land, don’t bother seeking information about modern cultivation techniques, and certainly don’t go about spreading any knowledge to others. Santosh exudes enthusiasm and professional pride in his fields.

In his 30s, Santosh is certainly much younger than his typical neighbors.  He holds an M.Sc. in agriculture, but so did several other large landholding farmers I met in MP, but they had none of the drive that Santosh displayed.

During my trip to Bihar and Orissa, I met several other enthusiastic farmers, who took pride in themselves as farmers.

In the Vaishali district of Bihar, when my colleague and I were visiting Farms-n-Farmers’  agri-advisory (extension services) centre, a farmer came almost running into the centre.  He asked us: “Exactly how many days before or after fertilizer application, should I apply the zinc?”

Unfortunately we had to burst his bubble of enthusiasm by sheepishly admitting that we have no clue; we were not the agri-scientists he had assumed us to be.  The agronomists were scheduled to arrive a few days later.  Determined to discuss the topic, the farmer sat down next to us and started a conversation with others in the shop.  He insisted on telling them that they should not make the mistake of applying zinc together with the fertilizers because it loses its potency; you have to do it before or after.  He just wanted to know exactly how much before, or after…

An Odisha farmer photographing an info-booklet using his mobile phone

In Daspalla district of Orissa, I met a farmer who came to the agri-advisory cum inputs retail centre of eKutir.  He picked up a booklet and started taking photos of it. Curious, I borrowed the magazine when he was done and discovered that the magazine had information on different varieties of vegetables (e.g. the various varieties of tomatoes).

Observing these farmers’ search for more information is heartening in a sector where such behaviour is usually quite rare.

The elusive progressive farmer

It is no wonder that all agri-sector social entrepreneurs I’ve met, start their work in a new area by first identifying one or two progressive farmers.

So who is this elusive species, the “progressive farmer”, that all agri-related social entrepreneurs seek?

The progressive farmer is typically young (often in his 30s), educated to some degree, actively seeks knowledge, is enthusiastic about farming as a profession, and, most important, willing to experiment with new ideas on his own field.

And, it is on the shoulders of such progressive farmers that the agriculture sector programs of most social entrepreneurs depend upon.

Paddy technology

Last week I visited CCD (Covenant Centre for Development), a Madurai based NGO, to observe their mango and medicinal herbs processing operations.  Along the way, I saw their “agro-service” operation. They have bought a machine (2.3Lakhs without subsidy) that does paddy transplanting.

Here are a few snaps showing the transplanting and machinery:

Paddy is grown in rectangular batches in a nursery
Paddy transplanting machine being loaded on to a truck
Recently planted paddy. Left-over planting material for use in case of plant mortality
Neat rows of machine planted paddy

Advantages of using the machine over traditional manual transplanting:

  1. Requires less planting material (and therefore seeds) per acre.
  2. Uniformity of spacing, rows, etc. makes it easier to apply a uniform quantity of pesticides later if needed
  3. No need to find and manage casual farm labour
  4. Lower cost compared to farm labor (at least in southern India)
  5. If the agro-service operation is professionally run, reliable and timely service can be a huge advantage

According to CCD, their single machine is so much in demand that farmers book it 18 days in advance and even pay a Rs. 1000 advance for booking.  CCD charges Rs. 2800 per acre and is able to cover about 1.5 acres per day.  With 2-3 people required for nursery cum machine operation, it appears to be a profitable business.

As they were describing their (very limited) operation, I was already wondering whether this could be viable as a stand-alone business? The answer seems to be yes. However, transplanting being a seasonal business, the question is how do you generate income in the off-season?  Are there complementary crops/machines that can be leveraged during non-paddy season?

Secondly, would such a business be attractive to rural youth?  And, could we (should we?) reduce the financial risk to the youth by leveraging different kinds of asset ownership/finance mechanisms, or possibly, franchise models?  Answers yet to be discovered…

A turkey that insisted on being photographed. Its role in the business is yet to be clarified!