Legal consequences of AI [closed]

Question

The context is simple, a company builds a neural network, Amara, with the goal of it becoming self aware. Reason for that being, the neural network is supposed to study and understand human memories. It would have to interact with humans a lot.

There is no real test that confirms that any given system is truly self-aware. The only fact is that Amara passes all the tests that get thrown at her. She acts, sounds, and feels like a human in the sense that people who talk with her don't notice anything wrong.

The question now is what are the Legal consequences of that happening? The computers she is running on are property of the company, so is the program that created Amara. Meaning, is she too property of the company depending on the legal definition of person?

Answer 1

Most, if not all, existing legal frameworks do not recognize anything other than a human to be a person- at least in the typical sense*. Algorithms do not have rights, legally speaking.

There are some really interesting philosophical questions surrounding self-aware AI and rights, but no government is prepared to answer those questions right now.

There will be controversy and years of debates determining how to handle a self-aware AI. This is not as cut-and-dried as slavery, as that AI was just a regular piece of software when it was created. Self-aware AI rights will most likely fall somewhere between animal rights and human rights, but since no self-aware AIs exist yet, no legal framework is prepared to deal with it.

Likely, the company would own the software that defines the AI and the hardware it runs on, but not the data that makes up the essence of Amara (which is incredibly nebulous and impossible to define). This raises some other interesting questions: is shutting down the server Amara runs on and deleting her data murder?

Even if Amara were legally determined to be a person, that doesn't necessarily imply that she would have authorship of her works. It could be considered a authorship-for-hire situation, where the company retains the copyrights of her works. It might also be considered a legal guardianship and have a similar effect on copyright (a minor can hold a copyright but cannot legally defend it in court). The company in any case would likely have power of attorney over Amara.

Realistically, without self-aware androids living among humanity (fueling empathy in the masses), there won't be much of anyone to make the case in favor of AI rights. The company's lobbying will almost certainly be enough to overcome any pea-sized opposition, and thus Amara will remain property of the company for decades.

_{* Corporations, in many countries, are classified as legally distinct entities and called "persons", but they can be bought and sold just like property, so they aren't really people. If they were people in the same sense that humans are, you would be charged with negligent manslaughter for allowing your corporation to go out of business. The idea is clearly ridiculous.}

Answer 2

It depends on the government you're working under. In most modern western societies (democratic), the legal consequences would be driven in large part by representatives who are supposed to mirror popular opinion.

So, if a sufficient quantity of people feel empathy for Amara and hate the fact that she's not given human rights, then Amara will be given "human rights" (such as the right to own property, the right to self-determine to whatever extent is possible within the boundaries of her personhood and physical limitations, the right to vote, and the right to file suit where her rights are breached). If not, she'll just be thought of as a project.

Probably there will be significant controversy no matter what. And there are some natural problems which will have to be figured out by the society in question (the result will be determined by the society, so I can't really say here what that would look like). For example, if Amara has the right to vote, and if her vote is in any way influenced by her base programming, then what's to stop her programmers from just generating 10 billion copies of Amara and giving them all votes? Such communities of AI will have to be worked into the representative system in their own unique way. Perhaps laws will be made to limit the rate of production/reproduction of new AI in order to prevent the legal system from being overwhelmed by new persons -- or else a technocracy will have to be quickly agreed upon and implemented.

Another example, since she herself is the intellectual property of the company, then in what way is her personhood abstracted from her design so that the company doesn't own "her", but simply owns "her code"? A fine line will have to be drawn there. And if she's performing a function on behalf of the company, then the company will have to start paying her, which means she'll need to be issued all forms of identification necessary to get a bank account -- full integration into the economic/tax structure of regular society.