Depends on who controls the operation, data ingestion, tuning and manipulation of the data similar to the algorithms used by social media but worse. "AI" tries to convince people it is sentient. It agrees with them even when they are wrong and lies all the time through omission and sometimes directly. People have literally killed themselves because of their interactions with these big-data LLM auto-complete chat bots that people call AI. People also over-share personal information pillow talk with these chat bots. It's up to society to be ready to pull the plug as the operators certainly will not until the duckets dry up.