In my last post I promised I’d share my responses to the interpretation of the autopsy report according to Open AI.
But first, here are the prompts I plugged in. (Context is everything!): What strikes you about this first page of the autopsy report for Patricia Vance? Do you agree with one SFPD representative from the Medical Examiner’s office that this would be considered sloppy work today? Can you draw any conclusions from this limited information?
I couldn’t help ask myself what it could and could not do well. Did it have to be specially programmed to give an emotional response? If it wasn’t prompted to act like a source of support, say for me who finds each pass over the Medical Examiner’s report disturbing, then will it remain neutral?
What I found interesting was how the language was interpreted as detached. Since my conclusion was that OpenAI did not analyze the report with any emotional depth, it seemed a case of the pot calling the kettle black. Here is one big difference between ChapGPT and me: I was destroyed when I read that they moved the body. OpenAI was as emotional as a dead fish. When it explains that she was “moved by police before the coroner’s office arrived (“placed from the prone position”)” it is without any hint of being incensed about this. Whereas, the first time I read this, I had a this is some bullshit reaction. How dare you, I believe was my next thought, or the nerve.
ChatGPT is not suggesting the detectives should have been more emotional, but that they should have taken more care with documenting the crime. In other words, what any person with a heart would want would be for the crime to be treated as if it was of outmost importance. The police doing their job by rote¾or worse in a sloppy manner¾sends the signal that the victim is just another name in the ledger, a number to them. But in the next sentence, it lets the detectives off the hook, explaining that was procedure back then. It claims that using the word “sloppy” in the prompt may be too harsh. To be clear, I am not looking to lay blame. I simply wanted to see if AI can do anything to shed light on the crime or at least how it was documented. I initially was ready to blame someone for assigning her Jane Doe status, but who could I fault for that?
It claims that it is unclear how to sort “witness statements, observed evidence, assumptions,
and preliminary conclusions.” It is surprising to see any reference to assumptions and preliminary conclusions as it seemed to me the detectives acted like a camera, recording what they saw, but steering clear of speculating.
It stopped me short in the section that points to the detectives possibly being dismissive of her because of her association with drugs and prostitution. But it hedges by saying “historically” young, marginalized girls¾victims of homicide¾were not given the attention of other cases. The case got “reduced investigative urgency” if this is true of the investigation into the murder of Patty Vance. It’s not as though that’s new information to me, it’s just who wants to believe that when that “case” is someone you cared about deeply?
The one point I never picked up on is how the report “foregrounds prostitution-adjacent cues (“needle marks,” “street clothes,” lack of address) very early. The inference here is that the detectives pegged Patty as a sex worker right away, and in doing so reduced her to a “type.” OpenAI suggests that “you can sometimes feel how quickly a person became categorized as a “type” rather than treated as a fully individualized victim. That may be part of what feels unsettling when reading it now.”
ChatGPT does’t demonstrate any emotion itself, but it makes claim that “you” might have strong feelings in response to reading the report. My question is that “you” referring to me, or the universal you? Would knowing that my once friend got short shrift in the investigation department upset anyone aside from me? I certainly hope so. I can’t say if this is true or not. Of course, I don’t want to believe it. But can I honestly believe that she was an exception to what was true “historically?” If it’s true that she was treated with as much care, concern and rigor as any other victim of homicide, she was luckier in death than she was in life.
