> In present reality, AIs don't do anything a human doesn't tell them to do. This seems unlikely to change. Not in my reality. I ask AIs to do things (write co

> In present reality, AIs don't do anything a human doesn't tell them to do. This seems unlikely to change. Not in my reality. I ask AIs to do things (write code, mainly), and they do. But sometimes they go ahead and do something I didn't ask them to do, like deleting a file. I only asked them to write code, and yet they decided on their own to delete a file. I think the distinction is that, in their current form, AIs have the agency that we give them (--dangerously-skip-permissions), and they use that agency in unexpected ways. A more correct statement would be: AIs can decide to do things on their own, within the limits of agency that they've been given. A moral imperative would be: With an understanding of how random we know an LLM's output can be, we should exercise extreme caution when we grant them agency. On the other hand, we can't all exercise "extreme" caution all the time. But we expect it of our civil engineers when they build bridges, and our architects when they build houses. Computer Science needed to become Computer Engineering some time ago, where it would have enforced the same kind of ethics training that the other established engineering professions require[0]. [0] https://en.wikipedia.org/wiki/David_Parnas#Technical_activism https://en.wikipedia.org/wiki/David_Parnas#Technical_activis...