Isaac Asimov's Laws of Robotics Need an Update for AI

This is a dumbass article.

First, it self-evidently doesn't understand what the laws of robotics actually are and how they interrelate. You can't just tack on an unrelated fourth one - the current three are a mutually dependent set.

And second, it self-evidently doesn't understand that current "AI" isn't actually intelligence in any way, shape or form and possesses no self-awareness of any form, and therefore cannot be made subject to laws at all.

Laws can only meaningfully be applied to the people who use the current "AIs," since they're the ones who actually possess agency and self-awareness.

If you want to argue for something to stem the tide of deepfakes, that's the thing to argue for - straightforward criminal penalties for the people who employ "AIs" to make them.

Fourth Law: A robot or AI must not deceive a human by impersonating a human being.

Eh, the robots are sentient and can request more information before making a decision.

Current "AI" is just LLMs. And current AI depends on the user to explain the environment. There is no way for the LLM to be able to verify any of the conditions that will define "morality." And the LLMs don't understand anything, even to the level of what is a fact or what is a number. It can be tricked easier than a baby.

Wasn't the whole point of Asimov's literature works to demonstrate that there is no such thing as a set of laws that would prevent AI related catastrophes and that the whole issue is much more complex than that ? Why even try to build on them ? Didn't the author read the books ? Or even just I, Robot, you know, the one mentioned in the article ?