The Taste Test: What AI Reveals About the Real Purpose of Education

On the collapse of stored knowledge, the one thing a machine can never do for you, and a hundred year old answer to AI in Education.

| "Education is not preparation for life; education is life itself." — John Dewey

In any given wine shop, there's easily hundreds of options. Which one do you choose? How do you know if it's good or not? Luckily, you have your phone in hand and AI is here. Take a picture and ask the genie: "is it good?" and within seconds it tells you everything: the region, the vintage, the producer's reputation, how critics scored it, what foods to pair it with, whether it's worth the price. A million data points, delivered instantly, and practically free.

But when you bring the bottle home, unscrew and examine the cork, pour a glass, smell and taste the wine...

At the instant, you encounter an irreducible zero-point — the summation of your immediate and present experience. No amount of data can substitute for your experience. The wine hits your palate, and something happens that belongs entirely to you: the surprise of an unexpected note, a memory of a dinner years ago, the simple judgment of whether you like it or not. AI told you everything about the wine. It cannot tell you what the wine is to you and for you.

The same can be said about any activity and any experience. How the sand scorches the skin between your toes on a hot summer day by the beach; how nostalgia and old heartbreak re-emerges from an old song; how you see your entire world held in the eyes of your partner or loved one.

Perhaps this sounds like trite or banal philosophy, but it's actually the most practical and important question in education right now. This is what I believe is the key to understanding what education should be in the age of Artificial Intelligence. Essentialized, this could very well be all there is to being human.

The Collapse of Stored Knowledge

For most of the history of formal education, a significant part of education was devoted to getting knowledge into people's heads. John Dewey, in Experience & Education, saw through the knowledge-transmission model and describes traditional education with an uncomfortable resemblance to today:

The main purpose or objective is to prepare the young for future responsibilities and for success in life, by means of acquisition of information. Since the subject-matter as well as standards of proper conduct are handed down from the past, the attitude of pupils must, upon the whole, be one of docility, receptivity, and obedience. Books, especially textbooks, are the chief representatives of the lore and wisdom of the past, while teachers are the organs through which pupils are brought into effective connection with the material.

The assumption was reasonable: retrieval was slow and expensive. A doctor who had memorized diagnostic criteria could act faster than one who had to consult a textbook. A sommelier who knew hundreds of regions and vintages had a genuine professional advantage. Knowledge stored in the mind was a form of capital and leverage.

Dewey names the problem:

The traditional scheme is, in essence, one of imposition from above and from outside. It imposes adult standards, subject-matter, and methods upon those who are only growing slowly towards maturity. That which is taught is thought of as essentially static. It is taught as a finished product, with little regard either to the ways in which it was originally built up or to changes that will surely occur in the future. It is to a large extent the cultural product of societies that assumed the future would be much like the past, and yet it is used as educational food in a society where change is the rule, not the exception.

This was written in 1938. An educational system designed for a world where the future resembles the past, deployed in a world where change is the rule. That was already a problem in the age of radio and automobiles. In the age of AI, it's absurd. If you had some knowledge about wines, you probably would've been fine without the help at the wine section. But now with a phone and internet, there's actually no need to know beforehand. AI has practically devalued all stored and storable knowledge overnight.

This is what a philosopher understood about the AI crisis in education — in 1938.

The Dead End Question (And What Schools Are For)

I think this is where the anxiety comes from. Not AI itself — but the dawning realization that the very thing school has been doing for centuries just stopped being necessary. Today's education, even at its most innovative, is still largely organized around filling you with knowledge that generally doesn't need to be memorized anymore.

The phone in your pocket can retrieve more information about wine than any sommelier who has ever lived. It can retrieve more about diagnostic medicine than any individual physician. It can retrieve, cross-reference, and synthesize information at a speed and scale beyond any human capacity. The instrumental value of stored knowledge — its usefulness as a tool for action — is rapidly approaching zero. (P.s., I want to make an important distinction between retrieval and knowing. But this is a subject for another piece.) It just turns out that retrieval is most of what we were teaching at school still.

This is the disruption that education is struggling to integrate. And most of the responses so far have been inadequate: ban AI, or embrace AI, or teach "AI literacy" as another technical subject within the same old walls at school. Banning AI from education will age about as well as banning Google from research. From this point of view, "how should we use AI in schools" starts to sound like a question asked by horse carriages about the validity of motorized vehicles on cobble roads — the point is missed entirely.

How AI should be used in schools is a dead end question. What's the value of horse carriages anymore, if at all? (There are still great uses of horses.) The real starting question, now that AI has taken over one of its primary historical functions: what can, should, ought schools actually be for?

The Irreducible Core

So here is the argument, stated plainly:

AI has made the knowledge-transmission model of education obsolete. What remains — the only thing that remains — is the part that was always the real point: the cultivation of experience itself. The capacity to attend. To discern. To judge. To sit with ambiguity and form your own relationship to what is in front of you. To taste the wine.

Education, Dewey argued, was never supposed to be about stockpiling information. It was supposed to be about the quality of experience — about cultivating people who are capable of richer, more discerning, more connected encounters with the world.

This is actually the hardest thing to teach. It is an essentialization — a stripping back to the real question of what it means to be human. Training someone's palate is harder than having them memorize a wine list. Teaching someone to read — really read, with sensitivity to tone, structure, and silence — is harder than teaching them to produce a summary. Developing judgment is harder than transmitting information. It takes more time, more skill, more patience, and more trust between teacher and student.

It is, if Dewey is right, the kind of education that always mattered. We just got distracted by the storage function because it was useful. AI has taken over that function, and in doing so, given us the opportunity to examine what really matters.

Not: what should students know to be successful in the future?

But: what should students be capable of experiencing?

This is the first in a series exploring the philosophy of education in the age of AI.