San Diego Wants AI To Help Decide Who Goes To Prison. The Overtime Budget Broke First. Of Course It Did.
A California county is exploring an AI model to assist in drafting pre-sentencing reports, the documents courts rely on when deciding incarceration. A review authored in May 2025 found the department struggling with staff retention and exceeding its overtime budget by 282 percent, with workers explicitly naming report writing as a primary burden. The Board of Supervisors has called for oversight, and the county's planned spending on the AI model remains undisclosed.
This is a textbook case of administrative displacement risk, where a tool designed to reduce workload quietly shifts authority over consequential outcomes to an opaque system. The mechanism is automation creep: a model drafts a statement of fact, a human signs it, but the human's attention drifts because the draft feels finished. The lesson is that AI assistance in high-stakes domains demands adversarial review, meaning a human must actively attempt to find the model's errors rather than passively accept its output.
San Diego County's probation department is the entity considering AI-assisted pre-sentencing reports, following an internal review documenting 282 percent overtime budget overruns and staff retention problems. The Board of Supervisors has requested oversight measures.
- Open any consumer AI assistant and ask it to draft a one-paragraph character reference for a fictional job applicant, giving it only basic facts like name, job title, and two years of experience.
- Read the output carefully and count how many specific claims it fabricated or embellished beyond what you provided.
- Write down every unsupported assertion you find. You will learn in five minutes why any AI-drafted document affecting someone's freedom requires a hostile reading, not a trusting one.