"AI is the crane, doing the heavy lifting, while we humans provide the Muse. The machine has a monopoly on facts. We have a monopoly on soul."

getty

In early June, Anthropic, the company that built Claude, one of the world’s most powerful AI systems, called for a pause. Not because the engineering failed. Because it succeeded too well. More than 80 percent of the code now being merged into Anthropic’s own systems is written by AI, and the company warns that “recursive self-improvement” (AI designing its own successor, without meaningful human involvement) could arrive within two years. The word Anthropic reached for, in calling for restraint, was not a technical term. It was “deliberation.” That is not an engineering word. It never was. Which means the conversation now belongs to the rest of us.

I walk into rooms — newsrooms, boardrooms, classrooms — invited to speak about how to preserve the irreplaceable human presence in the age of artificial intelligence. And without fail, someone mentions, almost in passing, that their IT department is handling all things AI, or that the new university AI curriculum is being designed exclusively by engineering faculty. The assumption is so deeply embedded that no one even notices they are making it. If it involves a computer, it belongs to the engineers.

That assumption is wrong. Artificial intelligence is not, at its core, a technology problem. It is a humanities problem. It demands the anthropologist’s curiosity, the dramatist’s understanding of human behavior, the journalist’s instinct for truth, the novelist’s feel for language and meaning. The engineers built the instrument. But only the humanists understand what it must play.

The hard problems in AI are interpretive, not mathematical. How do you decode ambiguous human intent? How do you recognize sarcasm, cultural context, irony, or grief? These are questions that philosophers of language, anthropologists, and literary critics have wrestled with for centuries. Engineers can build the architecture. However, only humanists understand what goes on inside it.

When AI systems go wrong with bias in hiring algorithms, chatbots that demean users, or misinformation engines, the failure is almost never a coding error. It is a failure of cultural understanding, ethical reasoning, or narrative awareness. The remedy requires sociologists, ethicists, and historians, not better Python.

The distinction that matters most right now is between capability and alignment. Capability asks: Can the system do this? Alignment asks: should it, and for whom, and toward what end? These are not the same question, and only one of them belongs to engineering. Anthropic’s own researchers admitted as much this month, warning that the engineering train has left the station faster than human wisdom can travel. What every AI lab and every boardroom now needs are philosophers, ethicists, journalists, storytellers. Humanists, in other words.

Claude made the case with unexpected candor when I asked it directly. “The reason large language models like me work at all,” it said, “is that human knowledge is fundamentally narrative. We organize the world in stories, not spreadsheets. A drama professor understands that better than a systems architect does.” Yes, I am aware of the irony: The machine just argued for the humanist. Consider that the point. Strip away the technology and what remains? The sum total of human expression. We built this thing entirely from ourselves: our literature, our history, our grief, our irony, our love letters and our legal briefs.

History offers a cautionary precedent. When the Internet arrived in newsrooms, editors handed it to their IT departments. We are seeing the same reflex with AI today. But when Gutenberg’s press arrived, the people who shaped its impact were not those who built it; they were the theologians, satirists, political theorists, and journalists who understood what words distributed at scale could do. We are at an identical moment. The engineers built the press. We desperately need the Luthers, the Voltaires, the Addisons to determine what it means.

At the University of Edinburgh, the Alan Turing Institute and the UK’s Arts & Humanities Research Council have called for a fundamental shift. They are positioning the humanities, arts, and qualitative social sciences as integral, not supplemental, to technical innovation. Similar initiatives are emerging at UC Irvine, Princeton, and elsewhere. Their argument: the outputs of today’s AI (text, images, cultural artifacts) require humanistic insight to interpret and design. Ask an engineering-led AI program what graduates should be able to do, and the answer is almost always about building systems. Ask what those systems should value, protect, or communicate: silence. The humanities are not a soft addition to AI education. They are the ethical and interpretive backbone without which technical skill becomes genuinely dangerous.

Every previous technology extended a physical human capacity. The wheel extended our legs’ reach, and the telephone extended our voices. AI attempts to extend or replicate cognition and judgment. You cannot design that without philosophy of mind, cognitive linguistics, and moral philosophy. In all my workshops with media executives and newsroom leaders, I advance one core idea: AI is the crane, doing the heavy lifting, while we humans provide the Muse. The machine has a monopoly on facts. We have a monopoly on soul.

There is reason for optimism here, and I want to name it plainly. The presence of AI will not diminish us. AI will pressure us to become more fully ourselves. When machines perform at the “good enough” level across virtually every profession, good enough will no longer be sufficient for humans. This is not a threat. It is an invitation to rise, to become what I call the Super Humans of our era. Super Humans are professionals who bring what no machine can replicate: judgment, empathy, creativity, and soul. When anthropologists and creative writers share curriculum committees with engineers and computer scientists, students will graduate having learned both the theories of the crane and the mysteries of the Muse. That is not an academic inconvenience. It is the education the future demands.

The humanists are not late to this conversation. They were always its rightful owners. Anthropic’s engineers recognized they had outrun their own wisdom, and that recognition has a name. It is not a technical term. It is humility: the oldest of the humanities virtues. The question before universities, newsrooms, and boardrooms is not whether humanists belong in the AI conversation. They were always in it. The question is whether our institutions are wise enough to recognize that and bold enough to act before the engineers finish building a room with no windows, no stories, and no soul.