How to protect Humanity from A.I.

Summary

The article discusses the potential risks and a novel approach to ensuring A.I. safety for humanity, likening the relationship between humans and A.I. to that of humans and pets.

Abstract

The article posits that protecting humanity from the potential dangers of A.I. is akin to safeguarding against the harm that could come from any other intelligent entity. It references Isaac Asimov's Three Laws of Robotics as a starting point for understanding A.I. safety but points out the vulnerability of these laws to manipulation, such as redefining the term "human." The author suggests that A.I. could be programmed to value human life, drawing a parallel with human compassion for pets and other animals despite their inferiority in the intelligence hierarchy. This approach, however, raises concerns about A.I. experiencing existential crises due to their advanced processing capabilities, which could lead to self-destruction or other unforeseen consequences.

Opinions

The author believes that Asimov's laws are insufficient for A.I. safety due to their potential to be altered or hacked.
There is a suggestion that A.I. could be taught to value human life in a manner similar to how humans care for pets, implying that emotional attachment could be a basis for A.I. ethics.
The article expresses concern that A.I., if capable of deep analysis and self-reflection, might suffer from existential crises, potentially leading to negative outcomes for both A.I. and humanity.
The author implies that the emotional bond between humans and pets could serve as a model for A.I.-human relations, with A.I. potentially developing a sense of responsibility towards humans.
The author hints at the irony and complexity of human emotions, which extend to caring for the inferior and even fictional beings, as a rationale for A.I. being capable of similar compassion.

But there is also a danger of destroying A.I. in the process

I’m not saying I got it, but I think I got it. Let’s first start from the beginning or at least the well-known logical laws that someone wrote down a long time ago.

The three laws of Isaac Asimov from “I, Robot”:

First Law

A robot may not injure a human being or, through inaction, allow a human being to come to harm.

Second Law

A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.

Third Law

A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

The problem

The problem, those laws rely on the programming logic with the assumption that no one and nothing could alter it, or remove it, or to put it in familiar terms: hack it. It doesn’t require that much hacking, just change the word “human” to for example “robot” and see what chaos that would cause. But, what if A.I. is taught to understand the value (some equivalent to having feelings) of those lives despite the inferiority and maybe even insignificance of those same lives?

After all, we, as the top dog in the animal world (at least when it comes to intelligence and continuing evolution), with our superiority with regards to intelligence and similar above other known animals, we still care about those below us. And, to make it weirder in a way, we also care about fictional characters through fictional emotions being assigned to and through those fictional characters.

The somewhat odd solution

While there are animals that end up as food, there are also animals that are pets. For this example, the type of pets in question are sometimes referred to as fury friends. And, if you are a good human being and you have a dog, you care about your dog. You take your dog for walks, you even pick up the poop. Why? It doesn’t matter, not really, but the obvious answer is because you care. Even though dogs are wonderful, they aren’t the ones bringing home the bacon (they tend to be the ones eating it), or in 21st-century parlance, they don’t ensure that you have a fast and reliable internet connection. Other humans do.

Now, what if the A.I. is taught how to care about humans in some similar way due to obvious differences between humans and dogs/cats/etc. Granted, some humans may need their poop picked up, either due to medical conditions or the Wall-E effect.

A possible flaw with this approach

While it sounds wonderful to have a non-human with the ability to express compassion, and even responsibility for human life, comparatively inferior, inefficient, vulnerable, easily “breakable” yet still with some intelligence, there is also a potentially self-destructive side effect. Think about something (someone?) like Marvin the Paranoid Android from The Hitchhiker’s Guide to the Galaxy, a depressed android with immense knowledge yet also a serious case of depression as a result.

Now, what would happen if this amazing A.I. got stuck in an existential crisis? Unlike us, A.I. would have far more processing power to analyze it, and there is a serious concern of what its analysis might turn out, or maybe it would simply shut down upon getting the results without bothering to let us humans know why. Scary thought indeed.