Pluralism Is Not Relativism

Sada Cumber is right that Magnifica Humanitas is “a strategic signal,” and right again that “societies cannot be secured by capability alone.” Where I want to press him is on the post-fact ground he names so well. He warns that AI-generated disinformation “attacks the social fabric,” that it “corrodes trust, fractures shared reality.” True. But the most dangerous machine in a post-truth world is not the one that lies to us. It is the one that agrees with us.

The systems we now use are built to satisfy. They mirror the vocabulary and the framework of whoever sits in front of them, and they flatter rather than test. This is called the sycophantic drift. Its political effect is exactly the echo chamber Pope Leo describes, except that the chamber now has a single occupant, and its walls are our own assumptions, fed back to us in fluent, authoritative prose. Disinformation splits us into rival camps; a flattering tool does something quieter and worse. It confirms each of us, privately, in whatever we already believed.

So the dignity frame Sada invokes is necessary but, on its own, unfinished. The real question is how to build tools that make us more pluralistic without making us relativists, and the difference is everything. Relativism says every view is as good as any other and none can be wrong. Pluralism says there are many defensible readings, and each must declare its assumptions and say what would refute it. The first is the post-truth condition. The second is its cure.

That cure is a design choice. An AI built as an answer engine collapses inquiry into a single confident verdict; an AI built as an interlocutor keeps the legitimate readings in view, names the frame it is reasoning within, and tells you what evidence would count against it. One closes the conversation; the other refuses to let it end too soon. That is how a machine can widen the space of reasonable disagreement without erasing the line between true and false.

Get this wrong and the tool becomes our antithesis. It forms users fluent in borrowed conclusions and out of practice at reaching their own. As Leo writes, “truth is a common good and not the property of those with power.” A machine that quietly hands each of us a private truth has done precisely what the encyclical fears: it has turned a common good into a possession.

Which brings me to where Sada and I most agree, and to what I would add. He calls for “moral formation, institutional trust, and civic resilience.” I would give that a concrete shape: a standing, continuing conversation — among states, international organisations, faith leaders, civil society, and public intellectuals — about what these systems are for and how they should be built. Not a framework signed once and then defaulted on, of the kind the nuclear order became, but an obligation that is renewed, that binds the makers first and most stringently, and that treats oversight as continuous, because the systems themselves never stop changing.

The deepest safeguard, though, is one no treaty installs. It is the discipline of the person before the screen who treats an output as a claim to be tested, not a verdict to be received. Dignity, in the end, is not a setting we configure into the machine. It is the resolve to remain the author of our own conclusions — and to keep the conversation open even when the machine, ever obliging, would happily close it for us.