# AI and the Ritual Society AI lowers the cost of producing rational-looking traces. That makes it easier to confuse complete documents, workflows, and retrospectives with real judgment. ## Metadata - HTML: https://glenzli.com/en/notes/ai-and-ritual-society/ - Markdown: https://glenzli.com/en/notes/ai-and-ritual-society.md - Collection: Notes - Language: en - Published: 2026-07-02 - Updated: 2026-07-14 - Tags: ai, society, judgment, workflow ## Content AI is changing many things. Whether it will make the world more rational is still unclear. What is already clear is more subtle: AI makes it easier for the world to perform rationality. In the past, many rational-looking artifacts took time to produce: reports, summaries, retrospectives, study notes, literature surveys, and meeting notes. Now models can generate them in volume. The format is complete, the tone is professional, the structure is clear, and sometimes the result looks more convincing than what a person would have written. So a question begins to surface: > When the cost of completing a ritual becomes low enough, will we become more likely to mistake the ritual for real ability? This is not mainly a question about whether AI is good or bad. It is a question about how society observes behavior, evaluates people, and rewards what it can see. ## Rituals Are Not the Problem Notes, records, verification, citations, processes, and reviews all have practical uses. They help people understand a subject, work together, check results, and take responsibility. The trouble begins when those actions detach from the things they were meant to serve. At that point, a tool starts to become a substitute. The mismatch can be very smooth. Someone has not understood the material, but the study notes are complete. A problem has not been solved, but the retrospective is thoughtful. Research has not moved forward, but the survey and report look polished. Risk may not have gone down at all, but the review process has been completed. Once ritual begins to take over real function, society asks less often: > Did this actually change anything? It becomes easier to ask: > Did it leave behind something presentable, reportable, and auditable? AI happens to lower the cost of producing those traces. ## The Threshold Becomes Ambiguous Many fields have barriers to entry. After AI, those barriers become harder to read. On the surface, a field seems to be protected by knowledge, experience, and professional competence. In practice, a large part of its outer wall is made of expression, format, jargon, procedure, and posture. You need to know how a paper is written, how a proposal is arranged, how a report is presented, and what kind of structure a review tends to like. You need to use the local vocabulary. You need to know what counts as a "research gap," a "closed risk loop," a "methodological lesson," or "long-term value." None of this is meaningless. A field needs shared language, and it needs norms for expression. The problem is that mastering these rituals is often mistaken for mastering the field itself. Worse, in some places ritual is no longer just an outer shell. People spend enormous effort learning how to make something look right, while asking less and less about the thing itself. Real experience can become hard to see if it does not arrive in the expected format. Empty language can pass smoothly as long as it is properly shaped. At that point, ritual does not merely hide real competence. It slowly consumes it. AI breaks this mixture apart. It is very good at completing rituals. Papers, reports, proposals, study notes, review comments, management retrospectives: it can produce the surface of all of them. For many people, this means the ritual threshold can be crossed very quickly. That does not mean they have mastered the field. It does mean that the old threshold used to maintain a professional appearance has suddenly become cheap. This makes many systems uneasy. Once the ritual barrier is lowered, the remaining question becomes harder to avoid: > Where does the real professionalism of a field actually live? The answer clearly lies beyond format, terminology, credentials, and process. It also lives in judgment, verification, responsibility, and experience. AI will flood the world with things that look professional. For that reason, it may also force real professionalism to step out from behind ritual and face a more direct examination. ## From Study to Everyday Life This change first shows up in learning and work. A topic has not yet been digested, but the notes and study plan are already complete. A problem remains unsolved, while its retrospective and process record continue to grow. The material looks more substantial; judgment does not necessarily grow with it. Academia and everyday life are pulled in as well. A paper can become more and more paper-like without the problem becoming better understood. AI can help plan travel, diet, exercise, and career choices too. The convenience is real, but a person may spend less time facing the question "What do I actually want?" and more time carrying out a life plan that merely looks reasonable. These changes do not need to be condemned too quickly. AI does lower the cost of starting many things. It helps people turn vague thoughts into something with shape. When giving something a finished shape becomes easy, appearance and substance are easier to confuse. Complete notes may not yet amount to understanding, and a meticulous plan may not belong to the person following it. What matters is whether any of it enters the thinking and action that follow. ## Reading Can Also Become Ritual So far, the argument has mostly been about producing things: writing them, arranging them, leaving behind a presentable trace. But reading can become ritual too. AI does not only make text easier to produce. It also makes it easier to hold up a sign that says, "I have read this." In the past, if someone had not read a piece of material, there was not much to hide behind. Now they can say: > I had AI summarize it. Sometimes that is honest help. A long text may be too long, the material too messy, the time too broken. AI can sketch the shape first, so a person knows where to enter. The trouble begins when that help becomes a softer kind of avoidance. Someone sees the outline, but does not let it change the question. They receive a shortened version, then continue to pour out what they were already going to say. AI has helped them read, but the reading has not really arrived anywhere. This is not the absence of words. Quite the opposite: the screen is full of them, with documents, notes, summaries, lists, answers, and knowledge-base entries. The problem is that the words pass in front of the eyes without settling in. They do not make a person pause, change course, or stop asking a question that has already been answered. So reading, too, becomes ritual. A document is no longer an entrance into the matter itself, but a ticket for speaking. A summary is no longer compressed understanding, but proof that the information has been "handled." This may be one of the quieter forms of illiteracy in the AI era: not the inability to see words, but the inability to be changed by them; not the absence of information, but information failing to reach action. ## Agent Evolution and Text Loops Move one step further into automation, and "agent evolution" becomes a useful example of this ritualization. The phrase sounds futuristic. An agent reads material, summarizes experience, reflects on failure, proposes a new strategy, and improves itself. None of this is worthless. Reading material, drafting plans, comparing paths, and proposing candidates are exactly the kinds of tasks AI can help with. But without real tasks, a clear way to judge success, a record of failure, and a cost limit, this kind of "evolution" can easily become a loop of text. After failure, it writes a retrospective. After the retrospective, it proposes a new plan. If the new plan fails, it writes an even more complete retrospective. It looks like learning. It may also be only the ritual of learning. Real evolution has to produce a change that can be tested in the next attempt. Failed approaches are dropped, useful experience survives, and errors do not merely find a home in a well-written reflection. Otherwise, "evolution" only teaches failure how to explain itself better. ## Visible Traces A ritual society rewards visible traces. Judgment, understanding, and whether anything has actually moved forward are hard to see directly. Reports, meetings, process records, and slide decks are easy to count. Everyday life follows the same pattern: how a person is actually living is hard to measure, while check-ins, photos, and summaries are right there on the screen. Once these traces become the target, people learn to optimize them. AI did not create that tendency, which has been present in modern organizations for a long time. It simply makes the traces cheaper, more polished, and harder to distinguish from real progress. That leads to a quiet but dangerous result: > The real problem remains unsolved, while the ritual around it becomes more and more complete. ## Judgment Cannot Be Outsourced The deepest temptation of AI may not be that it writes words or code for us. It is the suggestion that judgment can be handed away as well. Judgment is tiring. It asks us to set priorities and make tradeoffs. It also asks us to admit that some ideas merely sound reasonable, and that some failures contain nothing worth turning into a lesson. Sometimes what people want to automate is not just the work, but judgment itself. They want AI to define the goal, plan the path, verify the result, and carry the complexity, while they keep the benefit, the identity, and the right to sign off. But there is an awkward consequence. If a system can independently define goals, judge value, choose paths, accept results, and take responsibility, then the person who keeps only the signature is no longer a necessary part of the system. Once judgment has been handed away, retaining the signature does not automatically mean retaining control. That may be more worth worrying about than the familiar question of whether AI will replace people. ## Process Can Be More Than Ritual Of course, not every process is a ritual, and not every automation is empty motion. If a study session changes someone's understanding, a proof constrains the argument, or a security check actually reduces risk, the form is doing real work. The question is not whether a process exists, but whether it constrains the outcome. Did it change someone's understanding or choice? Did it leave evidence that can be checked? Did it reduce the cost of getting something wrong? If not, even the most complete and modern process may amount to little more than ritual. ## Back to the Matter at Hand AI can do more than make rituals prettier. It can also remove rituals that serve no purpose and return our attention to the matter at hand. It requires ordinary actions: after reading something, try to explain it in your own words; when handed a life plan, remove the parts that do not belong to you; after AI drafts a workflow or summary, keep the goals and tradeoffs in your own hands. AI can search, organize material, draft a first version, catch omissions, and take care of tedious execution. The problem is not that it does too much. The problem is whether we still remember to ask whether the work is worth doing and whether its output is worth keeping. Those decisions still require judgment, and judgment is exactly what a ritual society most easily gives up. ## The Ritual Engine The AI era will not automatically make society smarter. It may first make society better at performing intelligence. More reports, fuller processes, and better retrospectives do not necessarily mean that judgment has improved. That is not an original sin built into AI. In the hands of someone with a clear goal and a willingness to verify the result, it is a powerful tool. But in a society already inclined to reward ritual, it will amplify that inclination too. So perhaps the most important question in the AI era is not: > Can we make machines do more? It is: > Do we still know what is worth doing? If we know what is worth doing, AI remains a tool. If we do not, it will help us perform the ritual more beautifully than ever.