3 ms·
What might be better is have a large language model interview the officer and the whole thing recorded and transcribed.
by jcims 2y ago
What might be better is have a large language model interview the officer and the whole thing recorded and transcribed.
- bee_rider 2y agoSeems possible the LLM could lead the officer in unnecessary directions. Could a machine interview be replaced by a checklist or flowchart, I wonder? These LLM’s that can perform actions that look like conversing and thinking always seem to prompt the question: why did we do it that way? IMO the point of an interview is often not to just get the questions answered, but to get a second set of eyeballs with responsibility, agency, and the ability to blow the whistle, to scrutinize the series of events. Unless an AI is created with those abilities, I don’t see the point in having it interview people.
- jcims 2y agoGreat point. All it would take is some evidence of bias in the follow up questions to invalidate that approach completely.
- c-linkage 2y agoWhy stop there? Train ChatGPT on the legal code then feed live video into it. ChatGPT will then fine violators immediately for those crimes that are fineable. For arrestable crimes have it write the summons and then use text to speech so it can radio a police officer and dispatch to your location. The officer doesn't even have to know why they are arresting you, and it means they can be completely unaware of the legal code because the infraction was already detected and adjudicated.
- jrflowers 2y agoThe other day an LLM told me that the singer Charli XCX’s real name is “Professor Charles Xavier”