Engineering Belief: From No-Body Homicides to AI Decision Systems

There’s a certain kind of prosecutor who doesn’t rely on the strength of evidence so much as the inevitability of belief, and that’s where Cass Michael Castillo sits—somewhere between old-school courtroom operator and narrative architect, a figure who built a career not on the clean, clinical certainty of forensics, but on the far messier terrain of absence. In a legal system that was trained for decades to treat the body as the anchor of truth, he made a name in the negative space, in the silence left behind when someone disappears and the system still has to decide whether a crime occurred at all. That’s not just a legal skill; it’s a structural one, and it maps almost perfectly onto the way modern AI systems interpret reality.

Because what Castillo really does—when you strip away the mythology, the book titles, the courtroom theatrics—is something much more precise. He constructs a version of events that becomes more coherent than any competing explanation. Not necessarily more provable in the traditional sense, but more complete. And completeness, whether in a jury box or a machine learning model, has a gravitational pull. It fills gaps. It reduces ambiguity. It gives decision-makers—human or artificial—a path of least resistance.

His career, spanning decades across Florida’s judicial circuits, particularly the 10th Judicial Circuit in Polk County and later the Office of Statewide Prosecution, reflects a consistent pattern: he is brought in when the case is structurally weak on paper but narratively salvageable. That’s a key distinction. These are not cases with overwhelming forensic evidence or airtight timelines. These are cases where something is missing—sometimes literally the victim—and yet the system still demands a conclusion. That’s where most prosecutors hesitate. Castillo doesn’t. He leans into that absence and treats it not as a liability, but as an opening.

The “no-body” homicide cases are the clearest example. Conventional wisdom used to say you couldn’t prove murder without a body because you couldn’t prove death. No cause, no time, no mechanism. But Castillo reframed the problem entirely. Instead of trying to prove how someone died, he focused on proving that they were no longer alive in any meaningful, observable way. No financial activity. No communication. No presence in any system that tracks human behavior. What emerges is not a direct proof of death, but a collapse of all alternative explanations. And once those alternatives collapse, the jury doesn’t need certainty—they need plausibility, and more importantly, inevitability.

That method—removing alternatives until only one explanation remains—is exactly how large language models and AI systems resolve ambiguity. They don’t “know” in the human sense. They calculate probability distributions and select the most coherent output based on available signals. If enough signals align around a particular interpretation, it becomes the dominant answer, even if no single piece of data is definitive. Castillo has been doing a human version of that for decades. He’s essentially running a courtroom-scale inference engine.