> ## Content Index > Fetch the complete content index at: https://modalpathethics.com/llms.txt > Use this file to discover other available public pages before exploring further. # Show Me the 90 - URL: https://modalpathethics.com/show-me-the-90/ - Published: 2026-09-21T17:20:56.000Z - Updated: 2026-09-21T17:20:56.000Z - Description: The constitutional design window is closing. - Author: Aidan Lawson - Tags: Inner Apocalypse > [Ten percent chance](https://modalpathethics.com/applied-case-the-dog-gets-the-ball/) that sufficiently advanced artificial intelligence produces a catastrophe from which humanity does not recover. I am not asking you to accept that number. For the next few minutes, assume everyone warning about human extinction is **completely** wrong. > **We still have a timing problem.** --- ## The Other Ninety. A great deal of artificial-intelligence discussion is organized around the bad branch. - What if we lose control? - What if artificial intelligence becomes more capable than its operators? - What if the system deceives us? - What if the race outruns safety? - What if everyone dies? These are legitimate questions where the evidence warrants them. I have another one. > Suppose everyone **doesn't** die. What happens then? If someone gives me a ten-percent probability of extinction, there is another **ninety percent** sitting silently beside it. Show me the ninety. - Who owns the systems? - Who can replace them? - Who gets access to the records? - Who can challenge a classification? - Who preserves evidence against the institution using the model? - Who can terminate an emergency authority? - Who can remove the coordinator without destroying the coordination? - What happens when hospitals, governments, laboratories, infrastructure, businesses, schools, courts, militaries, and ordinary administrative systems have rebuilt enough of themselves around artificial intelligence that turning one system off becomes dangerous? *“Humanity survives”* does not answer any of that. **Survival** is an enormous possibility space. The entire argument of *The Inner Apocalypse* begins there. --- ## This Is Why Modal Path Ethics Is Moving So Fast, By the Way. On August 13, [I announced that Modal Path Ethics was slowing down.](https://modalpathethics.com/modal-path-ethics-is-slowing-down/) This was a tremendous success for approximately six days. The plan was one article a day. One argument. One direction. Finish the book. Preserve what remains of my hands. Then I canceled it. Things kept happening. Artificial intelligence, datacenters, institutions, research, politics, infrastructure, agents, labor, science, law. The publication queue kept moving ahead of the website. Waiting several weeks to release an analysis of a field changing underneath the analysis began to defeat the point of even doing it. So now there are days with four articles. This is not an attempt to achieve a world record in becoming annoying. It is a synchronization problem. A reader who encountered Modal Path Ethics through **Batman** on May 2 encountered a framework still learning what kinds of objects it could inspect. The book was published in June. By August, the inquiry had converged on sovereignty, coordination, institutional replacement, artificial intelligence, jurisdiction, correction, and the conditions under which a useful system becomes too important to challenge. By September, those paths had produced a post-sovereign constitutional project. The queue is full because the inquiry is ahead of the website. I am trying to close that distance. --- ## The Window Is Already Open. There is another reason. Artificial intelligence adoption is no longer waiting for philosophical agreement. Stanford's 2026 AI Index reports that 88 percent of surveyed organizations were using artificial intelligence in at least one business function in 2025, up from 78 percent the year before. Reported generative-AI use reached 79 percent. At the same time, deployment of AI agents remained in the single digits across nearly every business function. The survey data are self-reported and directional, which matters. The combination matters even more. - **Broad adoption.** - **Early agency.** That is a window. The systems are already entering institutions. The deeper reorganization is still beginning. The United States federal government shows the same motion from another angle. The Government Accountability Office found that selected agencies reported 571 AI use cases in 2023 and 1,110 in 2024\. Generative-AI use cases went from 32 to 282\. A later GAO review described agencies continuing to acquire AI capabilities while struggling with expertise, costs, procurement design, and lessons learned. Again: > The future has not finished arriving. It has started installing itself. That distinction is why **timing** matters. --- ## Eight to Twelve Months I have been using roughly **eight to twelve months** as my working horizon for getting the bones of an AI-ready constitutional architecture into usable shape. That is not an AGI prediction, an extinction countdown, or a claim that something *magical* happens next summer. This is a design deadline because institutional dependence can grow while everyone is still arguing about the capability forecast. The architecture needs to be public, attackable, revisable, testable, and concrete while replacement remains comparatively cheap. - Bounded last words. - Independent evidence paths. - Jurisdiction that cannot define its own outer boundary. - Review capable of producing an actual remedy. - Continuance loads accounted for during institutional replacement. - Authority able to die without taking its function with it. - Plural systems capable of coordinating without one successful coordinator acquiring title over the entire field. None of those becomes easier after dependence hardens. Some become much harder. --- ## The Election: the Wrong Clock. This is also why “we will deal with it during the next political cycle” is inadequate as a timing model. Elections matter. They remain one of the major ways a public can alter who exercises political authority. But artificial-intelligence integration does not wait between them. - A hospital changes software. - An agency signs a contract. - A laboratory changes its workflow. - A company retires an old internal system. - A model becomes the interface through which records are searched. - A standard becomes common. - Workers learn one tool. - The old competence disappears through attrition because nobody needs to practice it anymore. - Another institution connects. - Then another. No constitutional crisis has to be declared. No election has to be canceled. Nobody has to wear a crown. The replacement cost just keeps rising. This is the same problem Modal Path Ethics has been chasing since the beginning of *The Inner Apocalypse*: a system can acquire constitutional power because it works. The danger appears when carrying an important function becomes an argument for controlling the conditions under which that function may ever be carried differently. The current book runway states the corresponding demand directly: build coordinators strong enough to act while preserving a field strong enough to let them lose. An **election** can replace a politician. It cannot instantly reconstruct an abandoned technical capacity. It cannot automatically migrate ten years of records. It cannot create a competing provider overnight. It cannot recover an evidentiary route that every institution quietly stopped maintaining. It cannot make an indispensable dependency dispensable by voting harder at it. Those are different clocks. --- ## Assume They Are Wrong. So grant the optimistic case. Artificial intelligence does not kill us. There is no sudden machine takeover. There is no terminal escape. The catastrophic-risk estimates were inflated. Wonderful. We still need institutions capable of replacing successful systems. We still need correction paths that do not depend upon the object being corrected. We still need emergency powers with endings. We still need evidence that can escape the institution it may condemn. We still need infrastructure that can survive changing its mind. We still need a constitutional answer to intelligence becoming infrastructure. And if the catastrophic-risk people are right? The number really is ten? Then stop showing me the ten for a minute. > **Show me the 90.** Show me the branch where intelligence becomes extraordinary and nobody has to own everyone else because of it. Show me the hospital that can replace its system without abandoning its patients. Show me the government that can use superior coordination without making every later government dependent on the same coordinator. Show me the artificial intelligence that can become indispensable without becoming sovereign. Show me the institutions capable of correcting something smarter than any person inside them. Show me the future where humanity survives **and can still change its mind**. That future does not arrive automatically because extinction failed to happen. We have to build it. And the irritating thing about windows is that they are easiest to notice after they close.