When answers become cheap, what remains human?

There are certain phrases we inherit without ever really questioning them. “To err is human” is one of them. It has always sounded less like an observation and more like reassurance. A reminder that mistakes are inevitable since it encompasses a part of the human condition. Until AI began making them too. I found myself returning to that sentence more often than I expected. Not because AI was wrong. Humans have always been wrong. We misremember conversations, misjudge people and convince ourselves that certainty is just confidence in a nicer outfit. But watching a machine hallucinate legal cases  with complete confidence – as happened in the now well-known Mata v. Avianca case where lawyers unknowingly submitted fictitious citations generated by ChatGPT –  felt strangely familiar. If machines can make mistakes too, then perhaps mistakes were never what made us human in the first place.

That thought led me somewhere I hadn’t expected. Every time AI learns something we once believed belonged exclusively to us, we quietly move the boundary. First it was arithmetic, then chess, then writing then art. And now perhaps even mistakes. It makes me wonder if we have spent all these years defining humanity by exclusion by asking what machines cannot do instead of asking what humans actually are. Perhaps that is why every new technological leap feels strangely unsettling. It isn’t simply because AI is becoming more capable. It is because each capability quietly asks us to reconsider where we had drawn the boundary in the first place.

Maybe judgement feels different because it isn’t really about the answers. Most difficult decisions are not difficult because we lack information; they are difficult because information refuses to tell us what matters. A hiring decision is rarely just about matching a résumé to a job description. A diagnosis is not only about identifying the most probable illness. Two candidates can both deserve the role. Two treatment plans can both be reasonable. AI can generate recommendations for both. Someone still has to decide. That is already becoming visible across industries. Indian hospitals increasingly use AI-assisted diagnostic tools to flag diseases such as tuberculosis or identify abnormalities in medical scans, but the final diagnosis, the difficult conversation with a patient and the responsibility for treatment still rests with the doctor. AI may recognise patterns, but it cannot own their consequences. We often say AI will replace expertise. I’m beginning to think it will do something stranger. It will make expertise ordinary. Knowledge was once the competitive advantage. In an AI-first world, interpretation may become one instead. When everyone has access to intelligent answers, judgement quietly becomes the scarce resource. That is why the World Economic Forum lists analytical thinking, resilience and curiosity among the skills becoming increasingly valuable. As answers become abundant, they also become inexpensive. Judgement, however, remains scarce. And perhaps every technological revolution reminds us that when one human capability becomes ordinary, another quietly becomes invaluable.

But judgement does not appear out of nowhere. It begins much earlier, with curiosity. Psychologists describe something called the Google Effect – our tendency to remember where information lives instead of remembering the information itself. I wonder if AI is nudging us toward something similar. Not forgetting facts this time, but forgetting what it feels like to sit with uncertainty. To resist the first answer. To let a question remain unfinished long enough for a better one to appear. Socrates understood this long before search engines or language models existed. His method was never about collecting answers. It was about questioning assumptions until something deeper revealed itself. AI, by design, is trained to respond. Humans, at our best, know when not to stop asking.

The philosopher Hans-Georg Gadamer argued that understanding is never simply the accumulation of information. It emerges through interpretation – through context, conversation and lived experience. Meaning is not stored inside facts waiting to be retrieved; it is something we construct. Perhaps that is where uncertainty has gone. AI hasn’t removed it. It has relocated it. We are no longer uncertain about finding answers. We are uncertain about interpreting them, deciding which deserve our trust and accepting responsibility when they turn out to be wrong. The uncertainty has moved from information to meaning.

“To err is human.” Maybe that sentence was never really about mistakes. Maybe it was about what comes after them: the questioning, the reinterpretation and the responsibility of deciding what they mean. AI may eventually learn to imitate almost everything we do. What I hope it never teaches us is to stop wondering. Because if answers become effortless, then perhaps what remains uniquely human is not our ability to produce them, but our willingness to keep asking whether they were enough in the first place. Maybe what makes us human was never intelligence, creativity or even our capacity to make mistakes. Maybe it is our ability to interpret those mistakes, question their meaning and decide what comes next.