The Time We Thought Was Wasted
AI can compress production time, but it must not compress the time legal judgment needs to mature and be owned.
All professions of judgment have their slownesses, and all are ashamed of them. Working up a matter takes weeks, differential diagnosis demands its successive tests, a negotiating position is built through rounds. This duration was always treated as a cost: a necessary evil, imposed by the material, that one dreamed of reducing. Artificial intelligence grants that dream with magnificent brutality: what took a week takes an hour. And it is precisely this success that reveals a truth slowness had hidden: production time was not only production. It was also, clandestinely, the time in which thinking happened.
For here is what compression brings to light. While a lawyer drafted an opinion, they were not merely drafting: they discovered, in writing, the weaknesses of their reasoning; they stumbled on a joint that forced them back to the file; they let, between two working sessions, an intuition turn over in a corner of their mind. Drafting was the vehicle of a thought that would not have happened without it. We believed we were compressing a task of execution; we discover we were also compressing the laboratory where the decision was being worked out. The time we thought wasted was working.
We thought we were compressing production time. We were also compressing, without seeing it, the time in which decisions matured.
The work that thought
The idea that production thinks must be taken seriously, for it collides with our picture of intellectual work. We readily imagine thought as preceding execution: one reflects, then one drafts. The experience of all practitioners says the opposite: one reflects while drafting, and often only truly reflects there. Writing an analysis forces the choice of an order, and the order is already a thesis; formulating a warranty forces the fixing of its bounds, and the bounds are the real subject; summarizing a file forces a decision about what matters, and that decision is the heart of the trade. The hand thinks. Production tasks are machines for forcing choices that pure thought, left to itself, would have deferred indefinitely.
This phenomenon does not concern novices only, and that is what distinguishes it from the question of training. The most seasoned partner rereading a contract is not checking clauses: they are rethinking the deal, and it is the line-by-line crossing that brings back the point negotiated six months earlier, the inconsistency between two schedules, the risk no one had formulated. Take away the crossing, hand them the synthesis directly, and you have not spared them menial work: you have deprived them of the journey during which their expertise was being exercised. Expertise is not a stock one consults; it is an activity that needs a terrain, and the terrain was the work itself.
The hand thinks. Production tasks were machines for forcing choices that pure thought would have deferred.
The conclusion that arrives too soon
What happens when the conclusion arrives before the reflection? The psychology of decision has documented it for decades: an available conclusion becomes an anchor. The mind does not start from zero to verify it; it organizes itself around it, seeks what confirms it, neglects what disturbs it. Rereading is not redoing: whoever examines a ready-made analysis explores a path already traced, while whoever builds it explores the terrain. The alternatives that were not written are not weighed and discarded; they never appeared. The visible quality of the text masks the invisible narrowness of the exploration.
AI installs this regime at the scale of the whole trade. Every question receives its answer before having been turned over; every file arrives with its synthesis before having been crossed; every position presents itself drafted before having been weighed. The practitioner becomes the permanent evaluator of conclusions they did not form, and evaluation, however rigorous, lacks the virtues of formation: it controls a result, it does not make possibilities emerge. The risk is not that the answers will be wrong; they will be right more and more often. The risk is that judgment, deprived of the terrain where it was exercised, shrinks to a right of veto over thoughts come from elsewhere.
It will be objected that veto is already a power, and that sorting proposals is worth forming them. But evaluating and exploring do not cover the same space. Whoever evaluates rules on what is shown to them; whoever explores encounters what no one was showing. Yet the best professional decisions are rarely the choice of the best option presented: they are often options the crossing of the file made emerge, reformulations of the problem itself, side paths no synthesis contained because they did not yet exist. Sorting presupposes the map as given; judgment, at its best, redrew the map.
When the conclusion arrives before the reflection, reflection no longer builds. It ratifies, or it vetoes.
Manufactured urgency
To this inner transformation is added an outer pressure, more prosaic and just as powerful. Production time was also what justified deadlines: the client waited a week because the work took a week. When the work takes an hour, the wait becomes inexplicable. Why does this memorandum need three days, if the machine produces it before lunch? The question is already being asked in every organization, and it is formidable because it conflates two durations the old slowness fused: the time to produce the document and the time to mature the position. The first has just collapsed. The second has not moved, for it belongs to the material: weighing a risk, anticipating an adversary, letting an intuition rest demand today what they demanded yesterday.
But the second time has lost its alibi. Maturation was financed, so to speak, by production: no one asked for the week of reflection to be justified, since it merged with the week of drafting. Now separated, it appears as what it is, a chosen delay, and every chosen delay must defend itself. Facing a hurried client, competitors who deliver within the hour, management that measures throughput, defending a time that produces nothing visible is a difficult position. The natural slope is to abandon it: to let the deadline align with production time, that is, with almost nothing. The profession would gain in speed what it lost in deliberation, and no indicator would show the loss.
And this slope has a ratchet: concessions of speed are never taken back. The client who once received a draft within the day will never again understand the week; the deadline granted in one matter's urgency becomes the standard for all that follow. Every organization knows it from experience: one descends the ladder of deadlines step by step, one never climbs back up. This is why the question of time cannot be left to case-by-case arbitration, which will always settle it in the same direction. It must be posed once, as doctrine: which acts demand maturation, how much, and at what price. What has not been protected on principle will be ceded by slippage.
Maturation has lost the alibi of production. It will now have to defend itself as what it is: a chosen delay.
What duration made
To defend it, one must say precisely what duration made. Three things, at least. Distance first: a position reread cold is not the same position; the night that passes detaches the author from their text and gives them back a third party's gaze, which no immediate rereading accomplishes. Testing next: a thesis that held for a week in its author's mind, exposed to doubt, to imagined objections, to returns from the file, does not have the solidity of a thesis produced and adopted within the hour; duration is a stress test. Anticipation finally: seeing the adversary's moves coming, imagining hostile readings, playing the chess game of litigation demands slow simulations nothing replaces. Duration produced nothing; it eliminated. It was the filter through which bad ideas died in private rather than in public.
This is why the apparent equality between the document produced in an hour and the document produced in a week is an optical illusion. The two texts may be indistinguishable; the two positions are not. One went through the filter, the other did not. The difference shows in no comparison of documents; it shows only in time: at the first reversal, the first serious objection, the first hostile reading. It is a difference of the same family as the one separating the tested plan from the brilliant plan: invisible on paper, decisive on contact.
Duration produced nothing: it eliminated. Bad ideas died there in private rather than in public.
Taking back the time
The answer cannot be to slow production down, and this must be said as plainly as for every nostalgia: refusing speed would be both futile and unjustifiable. The answer is to dissociate what the old slowness conflated: produce fast, decide in due time. Nothing obliges the signature to follow the generation; nothing forbids organizing, between the document and the commitment, the delay the material demands. This requires making maturation explicit, since it is no longer automatic: cold-review moments written into the method, positions put to the test before being settled, an assumed distinction, toward clients too, between the date of the document and the date of the decision.
It is, at bottom, an opportunity disguised as a threat. The time AI frees is real, massive, and it has no natural destination: it will go where organizations direct it. It can dissolve into volume, every hour gained becoming an hour produced, and the profession will leave the decade faster and poorer in judgment. It can be reinvested where it always had its true value: not in the making of documents, but in the maturing of decisions. For the first time in their history, the professions of judgment can choose their relation to time instead of enduring it. The choice will not present itself twice, for habits of speed, once taken, do not come undone.
AI gives the professions their time back. It remains to decide whether it will finance haste or judgment, and that choice comes only once.
