Excavating AI

Fabulous essay by Kate Crawford and Trevor Paglen, uncovering the politics and biases embedded in the guge image databases that have been used for training machine learning software. Here’s how it begins:

You open up a database of pictures used to train artificial intelligence systems. At first, things seem straightforward. You’re met with thousands of images: apples and oranges, birds, dogs, horses, mountains, clouds, houses, and street signs. But as you probe further into the dataset, people begin to appear: cheerleaders, scuba divers, welders, Boy Scouts, fire walkers, and flower girls. Things get strange: A photograph of a woman smiling in a bikini is labeled a “slattern, slut, slovenly woman, trollop.” A young man drinking beer is categorized as an “alcoholic, alky, dipsomaniac, boozer, lush, soaker, souse.” A child wearing sunglasses is classified as a “failure, loser, non-starter, unsuccessful person.” You’re looking at the “person” category in a dataset called ImageNet, one of the most widely used training sets for machine learning.

Something is wrong with this picture.

Where did these images come from? Why were the people in the photos labeled this way? What sorts of politics are at work when pictures are paired with labels, and what are the implications when they are used to train technical systems?

In short, how did we get here?

The authors begin with a deceptively simple question: What work do images do in AI systems? What are computers meant to recognize in an image and what is misrecognised or even completely invisible? They examine the methods used for introducing images into computer systems and look at “how taxonomies order the foundational concepts that will become intelligible to a computer system”. Then they turn to the question of labeling: “how do humans tell computers which words will relate to a given image? And what is at stake in the way AI systems use these labels to classify humans, including by race, gender, emotions, ability, sexuality, and personality?” And finally, they turn to examine the purposes that computer vision is meant to serve in our society and interrogate the judgments, choices, and consequences of providing computers with these capacities.

This is a really insightful and sobering essay, based on extensive research.

Some time ago Crawford and Paglen created an experimental website — ImageNet Roulette — which enabled anyone to upload their photograph and then pulled up from the ImageNet database how the person would be classified based on their photograph. The site is now offline, but the Guardian journalist Julia Carrie Wong wrote an interesting article about it recently in the course of which she investigated how it would classify/describe her from her Guardian byline photo. Here’s what she found.

Interesting ne c’est pas? Remember, this is the technology underpinning facial recognition.

Do read the whole thing.

Creative wealth and moral bankruptcy

Tomorrow’s Observer column, which for some reason is online today:

In the parallel moral universe known as the tech industry, the MIT media lab was Valhalla. “The engineers, designers, scientists and physicians who constitute the two dozen research groups housed there,” burbled the Atlantic in a profile of what it called the Idea Factory, “work in what may be the world’s most interesting, most hyper-interdisciplinary thinktank.” It has apparently been responsible for a host of groundbreaking innovations including “the technology behind the Kindle and Guitar Hero” (I am not making this up) and its researchers “end up pollinating other projects with insights and ideas, within a hive of serendipitous collaboration”.

That was written in 2011. In the last two weeks, we have discovered that some of this groundbreaking work was funded by Jeffrey Epstein, the financial wizard who took his own life rather than face prosecution for sex trafficking and other crimes. It should be pointed out that most of those researchers were entirely unaware of who was funding their work and some of them have been very upset by learning the truth. Their distress is intensified by the discovery that their ignorance was not accidental…

Read on

MORE danah boyd’s Acceptance Speech (link in the post below) is worth reading in this context, because she worked for a time at the Media Lab.

Why the tech industry has to change

From danah boyd’s acceptance speech on being given the 2019 Barlow/Pioneer award:

“Move fast and break things” is an abomination if your goal is to create a healthy society. Taking short-cuts may be financially profitable in the short-term, but the cost to society is too great to be justified. In a healthy society, we accommodate differently abled people through accessibility standards, not because it’s financially prudent but because it’s the right thing to do. In a healthy society, we make certain that the vulnerable amongst us are not harassed into silence because that is not the value behind free speech. In a healthy society, we strategically design to increase social cohesion because binaries are machine logic not human logic.

The Great Reckoning is in front of us. How we respond to the calls for justice will shape the future of technology and society. We must hold accountable all who perpetuate, amplify, and enable hate, harm, and cruelty. But accountability without transformation is simply spectacle. We owe it to ourselves and to all of those who have been hurt to focus on the root of the problem. We also owe it to them to actively seek to not build certain technologies because the human cost is too great.

Google’s big move into ethics-theatre backfires.

This morning’s Observer column:

Given that the tech giants, which have been ethics-free zones from their foundations, owe their spectacular growth partly to the fact that they have, to date, been entirely untroubled either by legal regulation or scruples about exploiting taxation loopholes, this Damascene conversion is surely something to be welcomed, is it not? Ethics, after all, is concerned with the moral principles that affect how individuals make decisions and how they lead their lives.

That charitable thought is unlikely to survive even a cursory inspection of what is actually going on here. In an admirable dissection of the fourth of Google’s “principles” (“Be accountable to people”), for example, Prof David Watts reveals that, like almost all of these principles, it has the epistemological status of pocket lint or those exhortations to be kind to others one finds on evangelical websites. Does it mean accountable to “people” in general? Or just to Google’s people? Or to someone else’s people (like an independent regulator)? Answer comes there none from the code.

Warming to his task, Prof Watts continues: “If Google’s AI algorithms mistakenly conclude I am a terrorist and then pass this information on to national security agencies who use the information to arrest me, hold me incommunicado and interrogate me, will Google be accountable for its negligence or for contributing to my false imprisonment? How will it be accountable? If I am unhappy with Google’s version of accountability, to whom do I appeal for justice?”

Quite so. But then Google goes and doubles down on absurdity with its prestigious “advisory council” that “will consider some of Google’s most complex challenges that arise under our AI Principles, such as facial recognition and fairness in machine learning, providing diverse perspectives to inform our work”…

Read on

After I’d written the column, Google announced that it was dissolving its ethics advisory council. So we had to add this:

Postscript: Since this column was written, Google has announced that it is disbanding its ethics advisory council – the likely explanation is that the body collapsed under the weight of its own manifest absurdity.

That still leaves the cynical absurdity of Google’s AI ‘principles’ to be addressed, though.

Most Facebook users are entirely unmoved by the Cambridge Analytica scandal

Sad (and predictable) but true — from Reuters:

NEW YORK/SAN FRANCISCO (Reuters) – Most of Facebook’s U.S. users have remained loyal to the social network despite revelations that a political consultancy collected information about millions of accounts without owners’ permission, a Reuters/Ipsos poll released on Sunday showed.

The Reuters/Ipsos poll adds to other indications that Facebook has so far suffered no ill effects from the episode, other than a public relations headache.

The national online poll, conducted April 26-30, found that about half of Facebook’s American users said they had not recently changed the amount that they used the site, and another quarter said they were using it more.

The remaining quarter said that they were using it less recently, had stopped using it or deleted their account.

That means that the people using Facebook less were roughly balanced by those using it more, with no clear net loss or gain in use.

In a way, all this does is confirm the fact that the vast majority of our fellow-citizens is deaf to ethical considerations. We’ve seen this for the best part of a century in the UK, where the vast majority of the population read (and pay for) ethically-dubious and politically-biased tabloid newspapers.

Sweeping the Net for… [take your pick]

From Ron Deibert:

The LGBTQ news website, “Gay Today,” is blocked in Bahrain; the website for Greenpeace International is blocked in the UAE; a matrimonial dating website is censored in Afghanistan; all of the World Health Organization’s website, including sub-pages about HIV/AIDS information, is blocked in Kuwait; an entire category of websites labeled “Sex Education,” are all censored in Sudan; in Yemen, an armed faction, the Houthis, orders the country’s main ISP to block regional and news websites.

What’s the common denominator linking these examples of Internet censorship? All of them were undertaken using technology provided by the Canadian company, Netsweeper, Inc.

In a new Citizen Lab report published today, entitled Planet Netsweeper, we map the global proliferation of Netsweeper’s Internet filtering technology to 30 countries. We then focus our analysis on 10 countries with significant human rights, insecurity, or public policy issues in which Netsweeper systems are deployed on large consumer ISPs: Afghanistan, Bahrain, India, Kuwait, Pakistan, Qatar, Somalia, Sudan, UAE, and Yemen. The research was done using a combination of network measurement and in-country testing methods. One method involved scanning every one of the billions of IP addresses on the Internet to search for signatures we have developed for Netsweeper installations (think of it like an x-ray of the Internet).

National-level Internet censorship is a growing norm worldwide. It is also a big business opportunity for companies like Netsweeper. Netsweeper’s Internet filtering service works by dynamically categorizing Internet content, and then providing customers with options to choose categories they wish to block (e.g., “Matrimonial” in Afghanistan and “Sex Education” in Sudan). Customers can also create their own custom lists or add websites to categories of their own choosing.

Netsweeper markets its services to a wide range of clients, from institutions like libraries to large ISPs that control national-level Internet connectivity. Our report highlights problems with the latter, and specifically the problems that arise when Internet filtering services are sold to ISPs in authoritarian regimes, or countries facing insecurity, conflict, human rights abuses, or corruption. In these cases, Netsweeper’s services can easily be abused to help facilitate draconian controls on the public sphere by stifling access to information and freedom of expression.

While there are a few categories that some might consider non-controversial—e.g., filtering of pornography and spam—there are others that definitely are not. For example, Netsweeper offers a filtering category called “Alternative Lifestyles,” in which it appears mostly legitimate LGBTQ content is targeted for convenient blocking. In our testing, we found this category was selected in the United Arab Emirates and was preventing Internet users from accessing the websites of the Gay & Lesbian Alliance Against Defamation (http://www.glaad.org) and the International Foundation for Gender Education (http://www.ifge.org), among many others. This kind of censorship, facilitated by Netsweeper technology, is part of a larger pattern of systemic discrimination, violence, and other human rights abuses against LGBTQ individuals in many parts of the world.

According to the United Nations Guiding Principles on Business and Human Rights, all companies have responsibilities to evaluate and take measures to mitigate the negative human rights impacts of their services on an ongoing basis. Despite many years of reporting and numerous questions from journalists and academics, Netsweeper still fails to take this obligation seriously.

A pivotal moment

The resounding ‘Yes” vote in the Irish Referendum on changing the Constitution to allow same-sex marriage is a pivotal moment in the history of my beloved homeland. And in the history of the world too, in a small way, because this is the first occasion in which legal equality has been conferred on non-heterosexuals by a popular vote.

My private expectation was that it would be a narrowly positive vote, and that it would be decided by the urban/rural divide, with the electorates of Dublin, Cork and Galway voting overwhelmingly ‘Yes’ and most of the rural constituencies voting ‘No’. In the event I was completely wrong: only one constituency (Roscommon-South Leitrim) went negative, and that by a small margin. There was still an urban/rural divide, but it was much narrower than I had expected.

Referendum_cartoon

Cartoon by Martyn Turner in today’s Irish Times.

What it means (and what the Archbishop of Dublin, Diarmuid Martin, conceded) is that Irish society has finally turned the corner towards secularity. What’s astonishing, in some ways, is that it took so long, especially given how long the revelations about the hypocrisy and criminality of the Catholic church over child abuse have been in the public domain. The idea that this decrepit, decaying institution could pretend to be a guide to morals (not to mention politics) was laughable for decades, but it seems that it is only now that its bluff has finally been called.

In one way, it was bound to happen, for demographic reasons — or what marketing consultants call “biological leakage”, i.e. the remorseless tendency of older people to pass away. But that doesn’t lessen the sense of wonder that it has finally happened. As the Irish Times put it in its First Leader,

“the time when bishops could instruct the Irish people on how to vote has long gone. What we may not have appreciated until now is that being a young, networked society has political consequences that can overturn the cynical conventional wisdom about voting behaviour, turnout and engagement.

This is the first Irish electoral event in which young people have taken the lead and determined the outcome and it has been a bracing, refreshing experience. It had been visible on the streets for weeks in the Yes badges that became ubiquitous during the campaign but it had its most potent and poignant expression in the multitude of young emigrants who came home to vote on Friday. Here, in a single gesture, was all the pathos of separation and longing; an expression of solidarity and belonging; and an enduring loyalty to the nation that had so signally failed them. The tweets from those returning to vote for marriage equality were at once inspiring and heartbreaking, testimony to our failure and their promise.”

The campaign was fascinating because it was, as Noel Whelan put it in the Times, “the most extensive civic society campaign ever seen in Irish politics”. In that sense, it reminded one of the campaign that propelled Obama to the White House in 2008. The people who masterminded it — Brian Sheehan and Gráinne Healey — have shown themselves to be consummate, canny strategists who crafted a campaign that was deliberately open and conversational rather than confrontational. (The chosen theme was: “I’m Voting Yes, Ask Me Why?”)

For me, it was especially cheering to see that a long, lonely and exceedingly courageous campaign by a fellow Joycean, Senator David Norris, had finally born fruit. Writing in the Times today, he recalled the long and winding road “from criminal to equal citizen”:

I have been privileged in my life to follow a remarkable trajectory from being defined into criminality, challenging the criminal law, losing in the High Court and Supreme Courts, finally winning out by a margin of one vote in Europe, seeing the criminal law changed and then starting to build on this basis for human and civil rights for gay people.

Fifty years ago my first boyfriend said to me outside a Wimpy Bar on Burgh Quay: “I love you David but I can’t marry you.” I still remember that all these years later.

Go forward 10 years when, after a debate on decriminalisation, the late Mona Bean O’Cribben remarked vehemently to me: “This isn’t just about decriminalisation. You have a homosexual agenda. You won’t be satisfied until you have homosexual marriage.” I turned to her and said: “What a wonderful idea, thank you very much madam, have you got any other suggestions?”

But there is another, intangible but real, aspect to this vote. One of the strangely positive side-effects of the ‘Celtic Tiger’ years — when the Irish economy zoomed from sensible economic development to casino property-development insanity — was that my fellow citizens experienced for the first time what it was like to be seen as successful by the rest of the world. It was suddenly, as some of them observed at the time, “cool to be Irish”. All of which meant that the bust and the subsequent economic collapse had an even harsher psychic impact: it turned out that we had been kidding ourselves; that we had, as Frank McDonald (the great Irish Times journalist) used to say, “lost the run of ourselves”.

But one of the most unexpected byproducts of Friday’s vote is that we can be genuinely proud of ourselves, and for a reason infinitely better than fuelling a crazed property boom: for once, we did the right thing. Not a bad day’s work.

Quote of the Day

“Those are my principles. If you don’t like them,… well, I’ve got others.”

Groucho Marx

Funny that this quote should the one that came to mind when I was thinking about David Cameron.

The Kony video: an ethical virus?

My take on the Kony video.

According to YouTube, 60 hours of video material are uploaded to it every minute – an hour a second. In the midst of such abundance, how can anything get noticed? Attention is now the scarcest commodity in cyberspace – which explains why virality is so craved by those with things to sell or messages to transmit. In that sense, the most significant thing about the Kony video is that it represents the most successful exploitation of virality to date. But when you delve deeper, it turns out that its success owes something to network theory as well as to storytelling craft.

Many years ago, the Stanford sociologist Mark Granovetter published a seminal article in the American Journal of Sociology on the special role of “weak ties” in networks – links among people who are not closely bonded – as being critical for spreading ideas and for helping people join together for action.

An examination of the spread of the Kony video suggests that one weak tie in particular may have been critical in launching it to its present eminence. Her name is Oprah Winfrey and she tweeted: “Have watched the film. Had them on show last year” on 6 March, after which the graph of YouTube views of the video switches to the trajectory of a bat out of hell. Winfrey, it turns out, has 9.7 million followers on Twitter.

The problem with taking the tabloids

Amid the furore surrounding the Leveson Inquiry, one aspect of the affair is curiously absent: the role of the great British public in all of this. To illustrate this, consider the paradox that at a time when an increasing number of Sun journalists have been arrested on suspicion of making or facilitating corrupt payments to public officials, Rupert Murdoch launches the Sunday Sun to replace the late lamented News of the World — and it sells 3m copies on its launch day. So three million of our fellow-citizens went out and, of their own volition, paid good money to buy the thing.

And that, it seems to me, lies at the root of the problem. The underlying cause of the malfeasance at the Sun and other tabloid papers is that the tabloid market is an intensely competitive one. That’s why journalism in Britain can never be a ’profession’ — with all that implies in terms of standards, ethics and professional sanctions: it’s a trade grafted onto businesses operating in a fiercely competitive market. So journalists on tabloid newspapers are under intense pressure to come with ’stories’ that will give their paper a competitive edge.

But sleazy journalism wouldn’t give them such an edge if readers exercised some kind of moral or ethical judgement when choosing newspapers to buy. So the responsibility ultimately rests with consumers of the British tabloid product. If they genuinely abhorred the kinds of journalistic practices now being unearthed by Leveson, then the incentives to break or bend the law would be dramatically reduced. Bad behaviour would be punished. But what happens instead is that bad behaviour is rewarded — by increased circulation. (***See footnote)

So the great mystery is why consumers of journalistic products seem to be ethically neutered. I had a disturbing insight into this many years ago when visiting some friends of a friend. The couple in question were lovely, decent, unpretentious people in their mid-sixties from a working-class background. I noticed that they were readers of the Daily Express and asked if they were regular subscribers. They were. So, I asked, did it bother them that the paper they read every day was owned by a pornographer?

What was astonishing (to me) is that they were completely floored by the question — not in the sense that they didn’t have an answer, but in the sense that the question seemed, literally, meaningless to them. The idea that there might be an ethical dimension to their newspaper purchasing habit had clearly never crossed their minds. So there was an awkward silence and the conversation moved on. But as I saw the sales figures for the new Sunday Sun, memories of that conversation came flooding back. And as long as media ’consumption’ takes place in that ethical vacuum, then the problems being unearthed by Leveson will continue to plague us.

…………..

*Footnote: As far as I can remember, there has only been one occasion in recent history where bad behaviour was punished by readers — and that was when Liverpool readers boycotted the Sun after its disgraceful allegations about the behaviour of Liverpool football fans during the Hillsborough disaster.

My colleague Andrew Cupples points out that the Sun’s readership in Liverpool has never recovered from the paper’s coverage of the disaster. He pointed me at a Guardian story on the 20th anniversary of Hillsborough, which reads, in part:

The newspaper, which has a circulation of more than 3m nationally, sold just 8,000 copies in the area on the day of the memorial service at Anfield, which was attended by more than 30,000 people.

Inside the newspaper, still known as “The Scum” in Liverpool, “lifelong fan” David Wooding, the Whitehall editor, delivered a poignant tribute to the men, women and children who lost their lives. But for those who gathered at Anfield this week, it was far too little and far too late.

At the Albert pub, squeezed next to the ground, football scarves and Liverpool memorabilia cover the walls and ceiling. The entrance of the pub has a poster mocking the front page of the Sun’s notorious splash, which appeared a few days after the tragedy. The tabloid’s masthead appears to be dripping in blood. “The truth,” it reads. “96 dead. Hillsborough 15th April 1989. Don’t buy the Sun.”

Tommy Doran, who works at the Albert, remembers one regular reading the Sun in a corner of the pub. “I went over to him and said: ‘What’s that?’ and he went: ‘The Sun.’ I just ripped it up into pieces in front of him.” Like many others on Merseyside, Doran will never forgive the decision of then editor Kelvin MacKenzie to lead on 19 April 1989 with a story headlined “The Truth” that was anything but. In it, quoting unnamed police sources and a Tory MP, it claimed drunken Liverpool fans urinated on and picked the pockets of the dead, hampered rescue efforts and attacked policemen.

And the Wikipedia entry about the Hillsborough catastrophe claims that:

Many people in the Liverpool area continue to reject buying The Sun as a matter of principle, and the paper’s sales figures within Merseyside remain very poor. It is the only major newspaper not to have articles published on Liverpool’s official website. As of 2004, the average daily circulation of The Sun in Liverpool was just 12,000 copies a day. Some Liverpudlians refer to the paper as simply: The Scum.

This is interesting, but I suspect it’s the exception that proves the rule.