Although the smooth veneer of AI might gleam with a new-car sheen, the rough edges below the surface reveal its inherent inequalities. Madhumita Murgia, the first artificial intelligence editor for the Financial Times, writes in her probing Code Dependent: Living in the Shadow of AI that generative AI “is altering the very experience of being human.”
Murgia illustrates the ways that AI affects all areas of our society from health care and policing to business and education. “Our blindness to how AI systems work means we can’t properly comprehend when they go wrong or inflict harm—particularly on vulnerable people,” she writes. Drawing on deep research and interviews with individuals around the world, Murgia humanizes this claim by introducing readers to the people at risk, as well as to those endangering them.
Sama, a U.S. company that outsources digital work to East Africa, promises financial and social mobility to people living in poverty. Daniel, a South African migrant, was told he would be working with marketing content. But when he got to work, he discovered that his job was to spend hours viewing images of “human sacrifice, beheadings, hate speech and child abuse,” all for a salary of $2.20 per hour. Officially, he was flagging graphic and illegal content, but he was also training Facebook’s algorithms to identify this sort of content on their own. Daniel sued Sama and Meta, telling Murgia, “These companies are only interested in profit and not in the lives of the people whom they destroy.” His lawyer, Mercy, is even more pointed about the inequities of the AI industry: “‘All revolutions are built on the backs of slaves. So if AI is the next industrial revolution, then those who are working in AI training and moderation, they are the slaves for this revolution.”
Murgia also shows the promise of AI. In western India, Ashita, a doctor, uses an app called Qure.ai to help screen for diseases like tuberculosis in rural areas that lack access to health care, allowing her to deliver lifesaving care more quickly. Her usage data and communication with developers improved the software so that it could be rolled out more widely, and tuberculosis diagnosis and treatment shot up 35% in the region.
Code Dependent is full of such bracing, complicated stories throughout as it uncovers the perils and promise of AI.”