I’d go with weapons systems designed to remove humans from the decision. Tools that people use to approve medications or treatment without actually understanding what they’re approving. Cars that remove human judgement from uncertain circumstances. AI systems that make employment decisions to shield people from responsibility or legal culpability.

Basically any situation with real consequences where you’re taking a person out of the responsibility or decision making loop.

Also certain non LLM AI technologies for extracting information and patterns from interconnected data sets. Basically automated mass surveillance systems.