We are drunk on information technology. It is tempting to think that the ready availability of computers will solve the problems posed by people (or the shortage of people) when we're considering how to staff and resource the educational process.
The laudable aim of having computer terminals on every desk in primary schools spills over into the Utopian fantasy that the cost of teaching can be kept down by a judicious mix of technology and minimally-paid classroom supervisors.
At every educational level, the benefits of technological resourcing are cried up, and the spoken word regarded pityingly as a premodern tool of communication.
I write this in the uncomfortable awareness that there is a powerful technophobe streak in me, and that my capacity for managing information technology is enormously unimpressive.
But I hope that, in addition to this embarrassing personal agenda, there is a serious theoretical point to be explored about the nature of learning, even the nature of knowing.
When it is suggested that the classroom teacher can in significant measure be replaced by the computer - and this is not wild fantasy, but the kind of thing sober people close to government have said in print over the past year or so - there are matters I think ought to be raised about how the enterprise is being defined.
Information technology depends on binary systems: you progress by choosing between determinate options - yes or no. The sophistication with which such systems can adapt to complex discourse is extraordinary - and it may obscure for us some underlying assumptions.
The learner is thought of in such systems as occupying no particular place, as being a floating subject of choice, faced always with a pair of alternatives.
On the other side of the program, the possible or negotiable alternatives are prescribed in advance by its author. It is the binary element and the preselection of options that allows computers to operate with such rapidity; it is also what makes them notoriously capable of ludicrously disproportionate errors.
Beyond a certain point, the computer can't ask, "what are you asking?" It can clarify a command by checking alternatives, but it cannot clarify an interest.
Human conversation characteristically involves the chance of uncovering what concern lies behind a question; it can create idioms for continuing the exchange that either party began. It can relate an apparently abstract question to a specific agenda - personal, political, ideological or gender-based. Very seldom do binary structures give enough room for this kind of exchange.
And the question of who has chosen the negotiable alternatives and why, the question of the politics of information, is not easily phrased in this context.
To put it another way, anything we might call critical learning - learning how to ask, and how to examine and dismantle or reconstruct an answer - is not going to be easy if the main tool of learning is binary.
There is nothing mystical about this; I'm not speaking of some sort of human wisdom that goes beyond the laws of contradiction, only noting the place of time and exchange in learning.
Education as a fundamentally conversational business requires the interaction of subjects. How odd that we use the term "interactive" as a designation for certain sorts of technological programme - as if two free and unpredictable agents were involved.
I realise that this Socratic dream will feel pretty remote from the junior classroom in Abertillery or Hackney. But in environments of privation and stress, it is more than ever foolish to imagine that artificial intelligence can replace human - because here the "what are you asking?" issue can often be central; the politics will be important - the reasons for not asking or not understanding will be painfully present in the educator's mind.
Long before it is any use putting a child in front of a screen, the specific character of the child has to be grasped in terms of what does and what does not make them interested (in the fullest sense of the word) in learning - that is, what their subjective concern is and what their objective profit might be.
Before anything gets mortgaged to the promise of technology in the classroom, before it is assumed that the human educator can be in any way relegated in the learning process, we have to ask questions that I don't hear being asked all that often in the educational establishment.
Is it all finally about the transfer of items of information, preselected by management, and of determinate skills? Or is education still to be thought of as critical and - alarming word - emancipatory? Once we are clear about this, we will know where to locate the technological enterprise within the larger process.
But even the best tools don't emancipate. People do.
The Right Reverend Rowan Williams is the Bishop of Monmouth