The point being made by the Great Sage of Catholic Journalism is that human fallenness is as concrete as a sidewalk and as self-evident as a headline for those who have eyes to see and ears to hear the exterior world around them.
In the rush of progress since Chesterton’s death in 1936, we have managed to litter the hidden, interiorized nano world of binary digitization with confirmation of our brokenness as well. The case is made incontrovertibly by a July 24 report in the Wall Street Journal involving not, as some might suspect, the degradations of the global porn industry but rather the newest beacon of purported human advancement, ChatGPT and its artificial intelligence variations.
The vast quantum of conversation and rumination about AI has concerned its potentially apocalyptic effect on human labour. It has extended to fears about our ability to work out our own destiny as a species once AI assumes control. But the Wall Street Journal story reveals in horrifying detail the deplorable darkness required just to get our bright shiny chatbots to launch stage earlier this year.
The reportage starts with a trope from the anti-sweat shop campaigns of the past two decades: labourers desperate for work in economically oppressed locales — in this case, Nairobi, Kenya — being forced to undergo indignities gusting to cruelty for pittance wages. Then it wades unflinchingly into the spiritual sewers of the Internet where the waste product is human souls.
ChatGPT’s parent company, Open AI, hired contract workers for as low as $1.46 per hour “to review and categorize thousands of graphic text passages obtained online, and generated by AI itself. Many of the passages contained descriptions of violence, harassment, self-harm, rape, child sexual abuse and bestiality,” WSJ reporters Karen Hao and Deepa Seetharaman write.
The purpose was to train AI’s so-called large language model to suppress such content and create a “safety filter” so the evils of the world don’t spring unawares on the millions who would use ChatGPT and its successors. The Kenyan workers were obliged to categorize AI’s “scrapings” of the Internet into four categories of toxicity with the worst being child sex abuse content, then upward to incest, bestiality, rape, sexual trafficking and sexual slavery.
One worker had to read 15,000 posts a month of “detailed paragraphs about parents raping their children and children having sex with animals.” After six months, he was so traumatized that his family fell apart because of his anxiety and depression. Similar mental health devastation was so widespread among the Kenyan workers that the source contracting company withdrew from the project. “It’s just so unbelievably ugly,” said Mark Sears, founder of the firm involved.
Yet even Sears acknowledged to the WSJ reporters there is no feasible way to make artificial intelligence functionally available in ordinary life without inflicting the horrors of damnation on human workers. The whole goal is to ensure we don’t see it when we ask ChatGPT to help us with our school assignments or work research.
The Church has a call here to at least critique, if not frustrate, that goal. It exists on at least two levels. The first, obviously, is to speak out for those so horribly damaged by what amounts to a form of digital neo-colonialism where African workers serve Western progressive dream merchants. But the second, perhaps less obvious, is to evangelize the world to renewed understanding that simply slapping the labels “progress” or “modern” on a proclaimed “advancement” doesn’t negate the eternal truth that we remain the poor banished children of Eve.
We began in sin. We are not redeemed by binary concoctions of ones and zeros converted into large language model chatbots, which bring more darkness with whatever light they might shed. We are redeemed by one God, in Blessed Trinity, that is the Word, not on the front page of the newspaper, but made flesh.