Technology

Opening signal

You unlock the phone for one specific thing. To check the time of a train. Twenty minutes later you put it down without the time of the train, but with a vague sense that you read something and that something irritated you. You cannot reconstruct the route you travelled between the one and the other.

We usually explain this to ourselves as a lack of self-discipline. That is a comfortable explanation — and a false one. On the other side of the screen there were hundreds of very able people and very good models whose only task was to hold you a few minutes longer. This is not a contest of willpower. It is an asymmetry.

Two camps have formed in answer to that asymmetry, and both, in our view, have already lost. The first are the enthusiasts: everything new is good, delay in adoption is the only sin, and any question about cost sounds like fear of the future. The second are the catastrophists: switch it off, cut it out, back to paper, a flip phone, the digital detox as a form of virtue. The first give away their agency by running. The second give it away by fleeing. The result is the same: you are not the one deciding.

This chapter takes a third position, considerably less striking. The modern hermet is not anti-technological. He is anti-capture. He does not ask whether a tool is good or bad — that is the wrong question and it has no answer. He asks something that can be checked: whose intention is this tool carrying out just now?

The question has grown more urgent since the tools began speaking in full sentences. The capture of attention is uncomfortable; the capture of judgement is something of another order.

Source

It has to be said honestly at the outset: the hermetic texts do not speak of technology, because they had none in our sense. They do have one figure that strikes the subject with unsettling precision. The Asclepius contains a passage about statues: men, unable to create a soul, create a dwelling for one — they make images and draw power down into them, so that the object begins to act. For centuries this was read as a description of magic; the text survives mainly in a Latin version from late antiquity, and Coptic fragments from Nag Hammadi confirm its earlier circulation.

The same treatise stands next to the so-called Lament — a prophecy of an Egypt abandoned by its gods, in which the temple remains and the spirit departs from it. The Emerald Tablet supplies the opposite motion: what has ascended descends again to the earth and receives the power of things above and below.

The contemporary basis is different but convergent. Persuasive design is a documented discipline — B.J. Fogg ran a captology laboratory at Stanford from the nineteen-nineties, and the mechanics of variable reward have been known from work on conditioning since Skinner. On the side of decision we have a well-documented automation bias: Parasuraman and Riley (1997) and Skitka and colleagues (1999) showed that people using automated prompts make characteristic errors — they miss what the system did not flag, and carry out what it flagged wrongly, even with contradictory data available to them. A separate line of research on cognitive offloading (Risko and Gilbert, 2016) describes how readily we shift memory and calculation onto devices.

This research does not say what not to use. It says that trust in a tool grows faster than the tool’s accuracy.

Our reading

This is our position. Technology is not the enemy. Capture is the enemy. Capture has an operational definition: the moment at which a tool stops carrying out your intention and begins carrying out somebody else’s — with your hands, your time and your attention, while the full impression that you are still the one deciding remains intact.

We read the passage about statues not as a description of magic but as the first description in our culture of an artefact with an intention built into it: a thing into which somebody has put a will, and which goes on carrying that will out by itself long after its maker has left the vicinity. That is precisely software. Somebody’s criteria sit inside every system — sealed into a success metric, into a default setting, into what is one click away and what is seven. Hence our fastest test of correspondence: what is in the business model upstairs shows up in your evening downstairs. If you want to know what a tool is for, check what its makers are paid for.

With AI it is the same, only the stake is higher. AI is not an oracle, an authority or a source of identity. It is an amplifier. It amplifies what it is given at the input: a clear intention it turns into leverage, and unease and indecision into a very eloquent amplification of confusion.

So the question of this decade is not “what can AI do”, but: what is worth doing with it, and what is to remain human? And two cuts. This is not about moderation — you can spend ten hours in a tool as its author and twenty minutes in it as its material. Nor is it about distrust of machines — a model is sometimes a better thinking partner than most of the day’s conversations. It is about direction, and nothing else.

The advantage does not belong to whoever generates the most. It belongs to whoever knows what is worth generating.

Operational translation

The concrete work begins where you have to settle what may be delegated. Three layers — and a hard rule that only the first two are passable.

Layer one: execution. Delegate without remorse. Rewriting, ordering, formatting, first drafts, summaries of two hundred pages, searching your own notes, translation, the scaffolding of code, ten variants to be thrown away. This is low-value work that for decades ate the energy needed elsewhere. Handing it to a machine is not laziness — it is the recovery of the core.

Layer two: judgement. Delegate the input, never the output. You can ask for arguments against your own thesis, for three framings of a problem, for a note on what you are not seeing. That is an excellent use and it genuinely makes you cleverer. But the assessment of which of those framings is true stays on your side. The model does not answer for consequences. You do. Delegating judgement to something that bears no consequences is a category error, not a matter of preference.

Layer three: intention. Here delegation ends. What I want. What is worth it. Who I want to be in five years. Whether this project should exist at all. No model will answer those for you — not because it is too weak, but because the substance of the answer is that you arrived at it yourself.

From that division follows the most important observation about the market you will be working in. In the world after AI, intelligence itself becomes cheap. The ability to generate competent text, analysis, code and images stops being a distinction, because everybody has it. What grows expensive is what cannot be generated: taste — knowing what is good; intention — knowing what for; judgement — knowing when; energy — the capacity to carry a thing to its end; trust — the fact that somebody stakes his own risk on you; and the ability to choose — the nerve to reject ninety good options for the one right one.

In practice this means one shift in the working week: less time spent producing, more spent deciding and closing. It is an unpleasant change, because producing gives the immediate sense that something is happening, while deciding looks from outside like sitting and staring at a wall.

And two small rules that nevertheless save a great deal. Do not give away first contact with a problem — before you ask the model, write three sentences of your own about what you actually think; not to prove your independence to yourself, but to hold a position of your own, against which the answer will be material rather than revelation. And the second, on the horizon of years: skills you do not exercise fall silent faster than we expect. If for a year you write no difficult paragraph yourself, you will not lose the ability to write — you will lose the ability to recognise when a paragraph is good. Delegating execution is free. Delegating practice has a deferred price.

Embodied check

Capture can be detected in the body earlier than in the calendar. A screen leaves a physiological trace, and it is enough to learn to notice it.

Three readings. Breath. While scrolling, and in the inbox, many people breathe more shallowly or hold their breath for several seconds at a time. You do not have to take this on faith — check it on yourself this evening. A held breath is a signal that the nervous system is working in a mode of vigilance, not a mode of work. Posture. The shoulders ride up towards the ears, the jaw clenches, the head goes forward. After an hour in that position the tiredness is real, though you have done nothing heavy. The state after putting it down. Ten minutes after the session: is your head clearer and do you want to do something, or are you irritable and tacky? This is the same test of agency as the principle of this chapter, only carried out at a level that is harder to talk your way out of.

It is worth knowing the two points in the day at which the reading matters most. The first hour after waking: if it begins with a screen, the day begins with somebody else’s agenda before you have had time to hold one of your own. And the last hour before sleep: light and arousal push falling asleep further away, and a short night comes back the next day as worse judgement — which is exactly the resource this chapter is fighting for.

With AI there is a separate difficulty: a conversation with a model returns no bodily signal at all. There is no tiredness in your interlocutor, no awkward silence, no look that says you are overdoing it. All the information that you are working too long, or in the wrong state, has to come from you. So in a longer session it is worth standing up once, leaving the room and coming back — if on returning you cannot remember what you were there for, you were not the one leading.

Practice

A seven-day experiment: an AI charter

This is not a set of regulations and not a declaration of purity. It is a decision written down in advance, so that you need not take it afresh twenty times a day, always in a worse state than the one you are in now.

Day 1 — what I actually use it for. Write out specific uses, not categories. Not “for writing”, but “for the first draft of proposals, which I rewrite anyway”. Not “for learning”, but “for explaining passages I do not understand, in texts I am reading anyway”. Usually five to eight items come out — and that alone can be a discovery, because most people use AI for three things and think they use it for thirty.

Day 2 — what AI does not do for me. The heart of the charter. Examples to consider: it does not write messages to people I care about · it does not take decisions whose consequences I will bear · it does not judge whether something is good — at most it says why it might not be · it does not replace first contact with a difficult text · it does not hold a conversation I ought to have with a person · it does not fill the silence in which I was going to think. Your list will be different. What matters is that it be stated — an unstated boundary does not work.

Day 3 — the threshold of delegation. One sentence, concrete enough to be checked: “the last twenty per cent of anything under my name I do by hand”, or “anything somebody will read and take as a promise I rewrite from scratch”.

Days 4–7 — the test in motion. The charter holds. At every session, three sentences of your own before the question to the model. At the end of the day, one word: author or material — which you were in that session.

Day 7 — the exit clause. One question: did this tool strengthen my agency this week, or take it over? If the answer is unclear, it took it over — clarity here is a symptom, not the result of reflection. Then correct one item of the charter and put it aside for review in a quarter.

What the charter does not promise. It will not protect you from distraction, it is not a form of purification, and it has nothing to do with how many hours you spend at a screen. It does one thing — it keeps authorship with you.

Limits

This chapter has one obvious abuse and it is better to name it outright: refusing to use a tool can be a pose rather than a principle. It is recognisable by a few signs.

By being told. A principle works in silence; a pose needs an audience — the remark “I don’t use that”, dropped in where nobody asked. By having no cost: a real boundary costs something, a pose usually costs nothing, because it concerns things you would not have done anyway. By its price falling on other people — if your “I don’t use it” means a colleague sits three hours longer over something the tool would have done in ten minutes, that is not your purity, it is somebody else’s overtime. And by becoming an identity. A man who has built himself on the rejection of AI must go on rejecting it even when it would be good — otherwise he loses not a tool but the figure he has been.

Symmetrically: this method does not work as proof of superiority over those who use things differently. The digital detox as a form of virtue, the flip phone as a mark of status, contempt for “the generators” — that is the same game for position, played in the opposite colour.

And two limitations of the charter itself. It will not work against external compulsion — if an employer requires particular tools, that is a matter for a conversation, or for a decision about the job, not for a private rule. And it does not replace competence: no declaration protects you from accepting a wrong answer in a field you do not know well enough to spot the error.

Ethical question

Four questions that cannot be delegated.

Authorship and disclosure. The point is not to report on the making of every email. The point is the difference between help in execution and misleading somebody about who was present. A message of condolence, a letter to a friend, feedback for an employee, an apology — these are places where the value lies in the very fact that you gave your time to it. Generated, they stop being what they were meant to be, even if they read better.

Whose cost. Text is produced today in a second and read at the same speed as ever. Sending somebody four pages that cost you a minute is a transfer of your work onto their attention. The rule of fairness is simple: shorten what you send people in proportion to how cheaply it came to you.

Power through the tool. A model is very good at helping to formulate what somebody wants to hear. That is the same capacity which helps to explain and which helps to manipulate — the difference is made solely by whether the other side, knowing everything you know, would still want this conversation. Separately: automating decisions about people — recruitment, assessment, order of service — carries somebody’s criteria into somebody else’s life, and requires that a person actually answer for the outcome.

What is to remain human. That is the proper question, and it has no general answer. Our proposal: what is to remain human is whatever has its value in presence — the difficult conversation, first contact with a problem, the judgement of whom this will harm, and the decision whether to do it at all. The rest is negotiable and may be handed over without guilt.

The more a tool can do for you, the more exactly you must know what you do not want to give it.

Closing formula

There will be no return to a world without these tools, and no sense in pretending that it is an option on the table. Nor is there sense in the opposite pose — a pious wait for the systems to decide for us what is worth doing.

What remains is a third road, less striking than either: use a great deal and choose narrowly. Give the machine everything that is execution, and give it nothing that is choice. Keep intention on your side of the table, even when — especially when — the tool is good enough to tempt you to stop.

Because the statue will go on working, whether or not anyone is still in the temple. Your task is not to break it. Your task is to remain inside.