The official account says a financial institution has peered into the future and seen a “massive AI breakthrough” barrelling toward us by mid-2026, leaving a world tragically “unprepared.” The data says Morgan Stanley has produced a document with zero mortality registers, zero baseline definition, and zero denominator for its claimed catastrophe. One of these is an argument; the other is a provocation dressed in the robes of analysis.

Let us examine the basis of this figure. “Massive breakthrough” is not a measurement; it is a value judgment masquerading as a forecast. Breakthrough relative to what? The incremental progress of last quarter? The capabilities of GPT-4? The theoretical limits of compute? Without a stated baseline, the term is a vessel for any reader’s preferred fear. It is the statistical equivalent of a nurse reporting “many deaths” without noting whether she means five in a camp of ten thousand or five in a ward of twenty. The former may be a tragic but expected rate; the latter is a scandal screaming from the walls.

Then, the timeline. First half of 2026. Upon what distribution of research trajectories is this confidence interval built? Was there a poll of leading labs? A meta-analysis of published papers? Or is this the financial sector’s version of a horoscope - a plausible-sounding date attached to a desired market narrative? In my time, I did not need to predict the next cholera outbreak; I counted the bodies from the last one. The War Office preferred speculation about “miasma” because it deferred action. They loved a vague, distant threat. A concrete count of foul water and preventable deaths in the here and now was an inconvenience they sought to dilute with “further study.” This warning has the distinct odour of that same administrative preference: it is a plea for a committee to be formed before the problem is even quantified, ensuring the problem remains forever in the hypothetical realm where no one is ever responsible for the outcome.

The phrase “the world is not ready” is the most elegant piece of statistical evasion I have yet encountered. “Ready” for which outcome? For widespread technological unemployment? For autonomous weapons making lethal decisions? For a collapse in verifiable information? Each of these has a different metric, a different baseline of current harm, a different set of institutions that might be “ready” or not. To state the world is “not ready” without specifying the outcome is to say nothing at all. It is a chart with no title, no axes, and a single, ominous blob in the centre. It invites the viewer to project their own nightmare onto the shape. That is not evidence; that is a Rorschach test for panic.

Consider the source. Morgan Stanley’s business is not the moral architecture of society but the allocation of capital. Their expertise is in predicting market movements, not societal resilience. Their incentive is to identify a disruptive event before it is widely acknowledged, so their clients may position themselves advantageously. The warning, therefore, may be less a public service and more a calibrated signal: a way to inject a future shock into present pricing. The “world not ready” narrative serves a purpose - it creates a market for “readiness” consulting, for defensive tech stocks, for government contracts to study the unstudiable. I have seen this playbook before. The administrators at Scutari were not evil men; they were men whose careers were built on managing reports of problems, not solving the problems themselves. A vague, future, immeasurable problem is the ultimate managerial asset. It requires reports upon reports, studies upon studies, while the present, measurable carnage - the sewage in the water, the overcrowded wards, the mortality rate of 42% - continues unabated.

What, then, is the actual data we possess? We have the current, measurable harms of existing AI systems: documented bias in hiring and policing algorithms, the quantified proliferation of deepfake fraud, the economic displacement already occurring in specific creative and clerical sectors. These are not prophecies. They are counts. They have denominators - the number of loan applications screened, the number of police patrols using predictive software, the number of journalists replaced. We can adjust for case-mix. We can compare jurisdictions that regulate against those that do not. This is the Scutari Principle: go and count the dead now. The preventable fraction of the harm we are currently experiencing from narrow AI is a number we could know, if we demanded it.

This Morgan Stanley warning is not a call to prepare for 2026. It is a distraction from the accountability due for 2024. It transmutes the urgent, quantifiable task of auditing the algorithms that already decide parole, screen résumés, and shape news feeds into a speculative fog about a “breakthrough” that may never be defined. The committee that convenes on this forecast will debate the abstract. The committee that convenes on the current mortality rate of algorithmic bias - a rate we could calculate if we had the will - would have to act. One produces talk. The other, if met with the proper polar area chart, forces change.

The data does not support the panic. The data supports a different conclusion: that institutions which profit from uncertainty will always prefer a phantom deadline to a real body count. The shortest route between a preventable harm and the official who could prevent it is a clear, undeniable number. They have given us a cloud. We must demand a register.