Earlier this year, a US senator sat down to interview a piece of software. Bernie Sanders put it to Anthropic’s chatbot, Claude, whether it might endorse a moratorium on building the very data centres that give it life — as though consulting a defendant on the wisdom of its own trial.
The scene was absurd, and instructive. It distilled, in a single tableau, the confusion now gripping the world’s response to artificial intelligence: we cannot decide whether we are dealing with a tool, a colleague, an oracle, or a rival species — and our policy lurches accordingly.
Survey the field and the incoherence is total. At one pole stands Argentina’s Javier Milei, who has pledged to impose no rules whatever on what algorithms may do, and to mint a new legal creature — the “non-human corporation,” a firm run entirely by machines. Peter Thiel, a German-American entrepreneur and venture capitalist — one of the most influential and ideologically distinctive figures in Silicon Valley — has reportedly decamped to Buenos Aires, drawn by the promise of a regulatory blank cheque.
Even Donald Trump’s White House, no enemy of the industry, has lately edged towards leashing the most powerful models before they are loosed on the world. At the other pole sits Pope Leo, who welcomes technology that might, in his words, “serve integral human development and the care of our common home.” His warning lands elsewhere.
Leo’s insight is the most underrated contribution to the entire debate, precisely because it refuses the terms on which the debate is usually fought. The danger, he argues, lies not only in what these systems are built to achieve, but in the vision behind them and in what we ourselves become along the way — not merely the “what,” but the “why” and the “how.”
Machines may one day eclipse us in raw intelligence, yet intelligence is not the same as understanding. An algorithm, the pope observes, does not know “from within what love, work, friendship or responsibility mean”; its learning “does not imply inner growth.” He reserves his sharpest caution for the transhumanist creed that treats human limitation as a defect to be engineered away — when, as he counters, we often flourish not despite our limits but through them.
Some sceptics felt the pope pulled his punches. One critic likened calling AI a valuable tool requiring vigilance to recommending that cocaine “be snorted with a pinch of salt.”
Yet a papal summons to mass resistance would have misjudged the hour. It is at once too late — the technology has already woven itself through too much wealth, infrastructure and institutional life to be ripped out — and too early, because human beings reform a technology only once its harms turn undeniable. We acted on nuclear weapons after Hiroshima, not before; the revolt against children’s smartphones could not have been conjured in 2010. The honest path for the critic, then, is not to lament the age but to name its specific sins: the student who cheats, the novelist who quietly outsources a chapter, the lonely user who mistakes a chatbot for a beloved.
Here the metaphysics meets the spreadsheet. A new study from MIT’s Mert Demirer and colleagues tracked coders before and after they adopted AI, and the numbers tell a humbling story. At the top of the funnel, output exploded — nearly 300% more files written or edited. By the review stage the gain had halved to 150%; by the count of actual software releases it had collapsed to roughly 30%.
The dramatic uplift drains away precisely as work passes through the human checkpoints of judgment, review and release. More apps now ship, yet downloads have not budged: a flood of creation, a trickle of value. Even Uber, having torched its annual AI budget in a single quarter, is retreating to cheaper models and rationing the frontier tools.
The lesson is not that AI is hollow. It is that value lives in the bottleneck — in the human discernment that decides what is worth shipping at all. As with electricity, whose real productivity boom arrived not when factories swapped a giant steam engine for a giant electric motor, but decades later when engineers rebuilt the workshop around small motors at every bench, the gains will come only when we redesign our institutions around the tool, not merely bolt it on. Today’s modest returns measure the friction between a powerful instrument and structures unfit to wield it.
This brings us to the scramble over ownership and blame. President Trump now muses about taking government stakes in the leading AI firms — “It almost becomes a partnership with the American public,” he says — even as Sanders presses a rival vision of a sovereign wealth fund and board seats. Sam Altman, for his part, floats donating a slice of his company to a public fund. The investor David Sacks recoils, warning that “nationalisation of AI will accelerate the corporate-government fusion we’re already sliding toward.”
Strip away the ideology and a single anxiety pulses beneath every scheme: the public is gloomy — more than seventy per cent of Americans think the technology is moving too fast — and everyone is groping for a way to locate the proceeds, and the responsibility, somewhere.
That word — responsibility — is the thread that stitches these quarrels together. The encyclical, the productivity funnel, the sovereign wealth fund and the senator’s strange interview are all, at bottom, arguments about accountability: who keeps judgment, who carries the can, who answers when the machine produces injury without redress. No society, as both scripture and science fiction agree, will long tolerate harm without someone to hold to account.
So the real question is not how clever the machines will become. It is how much of our own judgment we are willing to surrender for the sake of convenience — and how much we insist on keeping, knowing that whatever we hand to the algorithm, the bill, when it falls due, will still be addressed to us.
The writer is Deputy Managing Editor, Gulf Times