> You can only tell it to do what you know how to express.

Unfortunately that doesn't prevent much of anything since our language is extremely expressive and AI will just make up anything it needs to in order to fill in the blanks. When it comes to things like oppressing a group of people, or maximizing profits by dynamically setting your prices using a guess of each individual's income you can accomplish what you set out to do even with huge margins or error. We already know that people will use this technology who simply will not care if they price out or imprison a few people who shouldn't have been so long as massive numbers of their enemies are locked up or their profits continue to climb.

I think it's a significantly lower barrier than employing people to watch feeds non stop or review every instance of motion or person-detection. It would be pretty straightforward for the camera maker to test and evaluate a handful of presets that ship with the cameras, and the current state of vision models is already pretty excellent at identifying things in a nuanced and flexible way in images and video.

You can ask it to report anything that looks different recently. That'll catch some new things without needing to understand them in advance.