The Answer Before the Question

Something strange has happened to the strangest event of my lifetime: it has become ordinary.

First there was AlphaGo. A machine did something that intelligent people had assured other intelligent people was still far away, and the public response was mostly a shrug. Go was a board game, the machine was a computer, and computers were already good at board games. The news moved on because very few of us knew what kind of news we were looking at.

Then came the language models. For a brief, cheerful season they were parlor tricks: impressive parrots, autocomplete in a tailored suit. The debate among intellectuals was whether they could really think, and a great many people were remarkably certain that they could not. Then the safety debate arrived, and the question quietly changed. Almost overnight, we stopped asking whether the machine could think and started asking whether it wanted to kill us.

A few months later, even the apocalypse had become background noise. Commentators who had spent the previous decade worrying that the internet was making the young stupid now warned that AI would make the young stupid. Politicians pointed ominously at China and Russia. And people who had never once lost sleep over the philosophy of mind began asking a chatbot what to cook for dinner.

We skipped the conversation in the middle. Not the one about dinner—that part is refreshingly practical. I mean the conversation about what it means for human beings to build something that can participate in the one activity we always assumed explained our special place in the universe.

The technology has moved in and put its feet on the coffee table. Its meaning is still standing on the porch.

A convenient panic

There is a comforting explanation for all this: technologies become ordinary because that is what successful technologies do. The internet, smartphones, television, comic books—each arrived to a chorus of warnings about rotted brains and ruined youth, and each was eventually absorbed into the furniture of daily life. In hindsight, the panic looks less like a serious diagnosis than like a small ceremonial dance we perform before plugging the thing in.

That history ought to make us skeptical of today's panic. Still, it does not quite settle the matter. A hundred false alarms are an argument against jumpy neighbors, not a guarantee that the hundred-and-first house is fireproof.

What troubles me is not simply that people are panicking. It is that the panic is so conveniently small.

Will AI take our jobs? Will it make students lazy? Will it hand ugly governments nastier weapons? These are real questions, and I do not mean to wave them away. They are also reassuringly familiar. They allow us to fret about what AI will do to the arrangements we already have, which neatly spares us from asking what those arrangements were ever for.

For most of history, that larger question could be postponed with a perfectly respectable answer: we cannot. We cannot feed everyone. We cannot make society fair. We cannot stop aging. We cannot cure every disease, untangle every complex system, or repair every mess we make in the process of keeping warm and fed. Our inability did more than limit us; it let us off the hook. It allowed us to live on polite terms with the gap between what we claimed to value and how we actually organized the world.

AI presses on that excuse.

Not because it will wave a silicon wand and fix human life by Tuesday. Anyone promising that is selling something, almost certainly on a monthly plan. The point is simpler and more uncomfortable: if machine intelligence becomes a general-purpose tool for solving problems, then a number of things that used to belong safely to the realm of impossibility may cross the border into the realm of choice.

And choice is where our alibis run out.

Suppose we could end hunger and did not. Suppose we could make our institutions genuinely fair and discovered we rather liked the old tilt. Suppose we could treat aging and hesitated—not because the treatment failed, but because too many arrangements had been quietly built on the assumption that people would have the manners to die on schedule. What would we say about ourselves then?

The opposite possibility is no less unsettling. Suppose we actually solve these things. What happens to a civilization when the ancient struggles that organized its days are taken off the calendar? When survival, scarcity, and the slow certainty of bodily decline stop handing us ready-made purposes we never had to think up for ourselves?

We are the heirs to a wildly successful mechanistic worldview. It taught us how to take a problem apart, find the moving pieces, and repair the mechanism. I have no quarrel with that achievement; modern life is not generally lived by candlelight, with childhood routinely ending in an early grave. The trouble is not mechanism. The trouble begins when mechanism becomes the only language we know how to speak.

A purely mechanistic view does more than leave the question of purpose blank. It tempts us to mistake the part of reality we can measure and adjust for the whole of reality. Under that light, love, care, longing, and hope start to look like private hobbies of the nervous system, while pain, grief, fear, and dread look like engineering defects awaiting a patch. Yet these are not two separate lists—one of pleasant features to keep, the other of bugs to clear from the next release. The second list grows straight out of the first. We hurt because things can be lost. We fear because we care what happens next. Music becomes nothing more than pressure waves in the air only when the description stops just short of the creature who hears a tune.

Perhaps machine minds would inherit that deafness. I do not know, and I do not need to settle the question of machine consciousness to see the hazard right in front of us. Human beings already have a bad habit of treating whatever resists measurement as disposable. Left to itself, an optimization project can strip away the very conditions under which anything matters at all, and then proudly hand you a chart showing that every metric has gone up and to the right.

Saying this does not make suffering sacred, or turn fear into an heirloom we ought to polish and preserve. It simply recognizes that love, care, hope, pain, and longing are braided together down at the root. We cannot go in with garden shears, lop off everything unpleasant, and assume the rest will keep blooming as before. Our ignorance about what makes a life worth living is not a license to delete. If anything, it is a strong argument for conservation.

This is also the point where my own religion has to mean more than a rule about intellectual curiosity. Keep looking was never an instruction to race to the end of the score so we could shut the piano lid. The looking belongs in the same family as love and care; it is a way of staying awake to what is here. Alan Watts liked to say that life is more like music than a journey toward a destination. The point of a piece of music is not to reach the final chord as quickly as possible—if it were, the best composers would be the ones who wrote only finales. We were supposed to sing and dance while the music played. A civilization that mastered every mechanism and forgot how to hear the tune would not have solved life. It would have missed the performance.

We have become a civilization of extraordinary means and threadbare ends.

AI multiplies the means. It cannot supply the missing end. A more capable machine can help us get what we ask for; it cannot tell us whether the thing was worth wanting. And if we try to outsource that question to the machine, we have not become wiser. We have merely hidden our indecision inside a very expensive box.

Forty-two

Douglas Adams saw the shape of this comedy decades before the first large language model ever hallucinated a footnote.

In The Hitchhiker's Guide to the Galaxy, the supercomputer Deep Thought spends seven and a half million years calculating the Answer to the Ultimate Question of Life, the Universe, and Everything. The answer, when it finally arrives, is 42. This turns out to be less than clarifying. As Deep Thought patiently points out to its disappointed creators, the trouble is that nobody ever knew what the question was.

It is difficult to read that scene now without feeling that Adams slipped out of his own satire and took a job in Silicon Valley. We have built a magnificent answer-machine before agreeing on the question. “AI changes everything” may turn out to be our civilization's 42: a grand, portentous announcement that leaves us none the wiser because we cannot say what we mean by everything, much less what it ought to be changed into.

Years later, Adams traveled around the world with the zoologist Mark Carwardine to visit species on the brink of extinction. The book that came out of those trips, Last Chance to See, is funny, sharp-eyed, and quietly heartbroken. First he invented a comic fable about a planet built to discover the Ultimate Question. Then he traveled across the actual planet, paying attention to parts of it before they disappeared.

I have no idea whether Adams meant those two books to speak to each other. I only know that once you set them side by side, they are hard to unsee. The answer. The question. The last chance to see.

That may be why the current nonchalance around AI unsettles me more than the prophecies of doom. Ordinary life carrying on is not, by itself, a moral failure. People have to get dinner on the table. They should be allowed to use a helpful tool without first passing an oral exam in metaphysics; civilization would grind to a halt if every light switch required a philosophical seminar.

Still, familiarity is a potent anesthetic. We stop looking at a thing precisely because it works. We argue about its settings, its corporate landlords, and its effect on quarterly productivity, and somehow fail to notice the oldest questions in human life stepping quietly out from behind the server racks.

If fairness becomes possible, will we actually want it? If hunger becomes solvable, will we bother to solve it? If aging becomes negotiable, who gets a seat at the table? And if, against all historical habit, we manage to succeed—what do we do with ourselves on the morning after?

No machine can answer those questions for us. What the machine does, by its sheer competence, is make our favorite excuses harder to keep using.

We may still look at the answers and decide we do not care for them. Wait much longer, though, and we will have a hard time pretending the questions never came up.

---

Sources

• DeepMind, AlphaGo, official research retrospective.
• Douglas Adams, The Hitchhiker's Guide to the Galaxy.
• Douglas Adams and Mark Carwardine, Last Chance to See; BBC archive.
• Alan Watts, the spoken passage commonly circulated as Life and Music; original recording unidentified in our search.