This isn’t your usual artificial intelligence (AI) story—the one going gaga over the tech genius of the likes of Sam Altman or the one painting the technology itself as either revolutionary or pure evil. Instead, Madhumita Murgia’s Code Dependent explores the gray, murkier areas, putting the common folk and their stories into the spotlight. She tells the story of data labellers of Kenya, on whose backs the algorithms of self-driving cars are trained, and of a mother from the Netherlands who one day finds her child added to a list of potential criminals, a list created not by humans but a machine with all its racial bias. She tells the story of thousands of gig workers working for delivery apps, often underpaid, but also nobody to turn to to register their grievances, because, well, a machine is their boss.
“The point here was not to avoid Silicon Valley altogether, but to go beyond it, to uncover new voices, and bring forth how AI is already changing lives; to bring some urgency as people tend to care about things when it starts impacting them,” says Murgia, who was in India recently.
Some aspects of the book are troubling, some even harrowing, some bring hope, while some read Kafkaesque. A glaring aspect is the global North-South divide—the sheer income inequality around AI development, where the bigwigs of Silicon Valley are paid hefty sums while the mundane-repetitive task of training and deploying the tech using real-world data is outsourced to the poor populace of the developing world. Murgia writes about “data colonialism” in more places than one, and reading through the book, one cannot help but draw the parallels between the likes of the East India Company and the Big Techs of today.
“My thing here was to bring out such stories that would surprise people,” says Murgia, and successfully she does, as she brings forth how AI is already impacting the various facets of our lives—from livelihood, health, and social security to even personal freedom.
In a conversation with FE, she speaks about her book, advancements in AI, and how it is being the first AI editor at Financial Times:
Your book isn’t about ChatGPT but about the everyday AI algorithms. Was the book in the works even before the launch of ChatGPT? Was there a particular story or instance that steered you to write this book?
Much of my book was already done before ChatGPT’s launch. It stemmed from the story of Karl Ricanek, a facial recognition researcher. I interviewed him for a different story, and he spoke about his moral dilemma. He was proud of the work he had done as an engineer and felt that the technology could be a force for good, used to find missing children and catch criminals. However, in his lifetime, he saw it being twisted, particularly in the US, where police used it in live situations to identify people and at times against people of his own community. I found this to be an interesting tension and felt I wanted to write a book on such areas, where it isn’t about all the good stuff or even the bad stuff but everything in between—the messiness, the gray areas.
You write vividly about how AI and its development are already impacting people. When there is already such a detrimental impact, why isn’t anything done about it? Have the governments been slow to act?
It’s hard to blame the governments for being slow to act, as the technology has advanced rapidly and reached far and deep—in healthcare, jobs, public services, and criminal book—and it has reached all of these places because of the companies that are building these and the monopolies that they have.
I mention the concept of data colonialism. In action, it is because everyone has become so reliant on a few companies for their tech infra that we haven’t questioned the impact enough. But that is happening now. There’s a lot more awareness now, such as the negative impacts of social media and how algorithms work to create bubbles, so there is greater participation now from governments and civil society.
You mention data colonialism in several places. How do you define it? Do you see any parallels between the erstwhile East India Company and the Big Techs of today?
Data colonialism isn’t a term given by me. It was proposed by academics Nick Couldry and Ulises A Mejias, who wrote The Costs of Connection. For me, they made a very convincing case of drawing parallels with the private institutions that were part of the British empire. That has been happening with tech now. There are just a handful of companies not just holding the infrastructure but also the data needed to train these systems, and they have the know-how and expertise to build AI systems.
At the same time, there is less and less active academia left, at least in the US and UK. Also, a lot of the tech infrastructure that the governments use is by these companies. So you can see why they drew these parallels.
While AI has had detrimental impacts, it has also led to wealth creation in impoverished regions, where people now have newer avenues to make money working AI jobs. Has that stopped governments there from taking steps against exploitation and ill effects?
There is a lot of reporting on Kenya, the Philippines, etc, where there are such factories. I wanted to understand this better, so I went there and spent time in the homes of these data labellers to get a sense of the impact. Many spoke about how they could send their children to schools or get healthcare for their parents, which they could otherwise not afford. And these are digital jobs, better than physical labour in unsafe environments. But yes, the reason why there hasn’t been enough scrutiny is that governments see that these companies are employing people, creating new jobs, and injecting digital jobs into the economy.
The whole point of AI is to bring benefits, access and generate wealth for everyone. But that hasn’t happened in the Global South, where still it’s a minimum-wage job. The workers have to sign non-disclosure agreements (NDAs), hence, cannot talk to anyone about this, can’t unionise, and at times, don’t even know who their employer is.
Is technology advancing rapidly, or is it more hype?
Well, what the hype has done is that it has made the tech percolate into several industries. It’s now the priority of board-level members and CEOs of companies from pharmaceuticals to retail. It has made everyone feel some level of FOMO (fear of missing out), and feel that they need to get to the top of this technology, or else they’ll be left behind.
But AI itself, in terms of evolution, is evolving and becoming much more sophisticated at a much quicker pace than five years ago.
How is it working as the first AI editor?
The decision wasn’t driven by the launch of ChatGPT. FT’s editor-in-chief Roula Khalaf and I had discussed this before that. We had seen the rise of this tech, and it becoming not only more sophisticated but also more central to the tech industry itself, to the business models of Google, Microsoft, etc. And in the past two years, it has been sweeping every aspect of the economy, from HR to white-collar jobs and from office workers to creators and psychologists. So it became clear that it’s going to be a driving force of the economic story as well as the tech industry, and so the decision was made to be the leaders in breaking that story because we saw things were coming.
My job is threefold. First, to explain the rapidly changing technology itself and help people keep up with it, the latest breakthroughs and developments. Second, to look at the tech industry, which consists of some of the most valuable companies in the world, and how their dynamics have changed because of AI —the competitiveness, the stories there, and the new startups coming up. And the third is to look at other industries and help correspondents in those areas to shape their coverage. So most of the stories that have AI in them, wherever they are in the newsroom, will at some point come through me to give a sense—if this is a story or hype, is this interesting—and it helps us to be authoritative.
Book: Code Dependent: Living in the Shadow of AI
Author: Madhumita Murgia
Publisher: Pan Macmillan
Pp 320, Rs 699