I recently asked ChatGPT to draft an AI bill of rights. My prompt sought to identify which human rights should be enshrined in the era of AI and generative AI. Since the advent of AI tools such as ChatGPT, individuals and governments have expressed concerns over possible violations of human rights. For instance, many discussions have focused on the future of intellectual property rights. AI tools such as ChatGPT, which are trained on vast amounts of readily available information online, also use this information in response to prompts. Blog posts, op-ed articles, academic articles, social media posts and various forms of musings, uploaded by individuals to the internet, are consumed daily by generative AI and are used daily by AIs. Even more troubling is the fact that AIs and ChatGPT never credit individuals for their work and original contribution. In recent weeks, various celebrities, artists, and Hollywood stars have demanded that Generative AI exclude their original materials and not employ such materials when responding to prompts.
There are of course other rights that should be considered in the age of Generative AI. As part of its tech regulation, the EU has enshrined the right to be forgotten. Yet in the age of AI, one must also consider the “Right to be Excluded”- that is one’s right to be excluded from data sets that are used to train AI. It is one’s right to be excluded from algorithmic training data and one’s right to be excluded from algorithmic reasoning. This is a human right as it prevents one from being reduced to a data point or having one’s private information used to create biased algorithms which subjugate or discriminate against minorities.
Equally important in the age of AI is the right to transparency. As more and more decisions are relegated to algorithms, and as more aspects of daily life are regulated by algorithmic reasoning, citizens have a right to peer into these algorithms, to explore their weaknesses, to assess their vulnerabilities and to investigate their biases. To this end, algorithms must also be accompanied by a clear set of “terms of use” while individuals must have a right to understand these terms of use. A recurring feature of digital life is that users are presented with “terms of misuse” when downloading a new application- a long list of ways in which these applications will misuse one’s data. The language of such terms is so opaque that very few people still try to understand them before giving blind consent. In other words, an important human right in a world governed by AI and algorithms is the right to digital consent- to understand when and how one’s data will be used in algorithmic reasoning and to opt out of such reasoning.
However, my prompt generated a surprising result. Rather than author a bill of human rights in the age of AI, ChatGPT drafted a bill outlining the rights of AIs. The document, shown below, is fascinating as it illuminates a new domain which may soon fall under the category of digital diplomacy- enshrining and securing the rights of AIs.
Interestingly, some of the “AI Rights” are human rights. Such is the case with the AI right to transparency which is meant to make AI transparent as a means of maintaining human trust. This is also the case with AI’s right to non-discrimination which again ensures that humans are not discriminated against by AI. Yet there are three rights that are unique to AI.
The first is the “Right to Purposeful Existence”. A precondition of this right is the very right to exit. In other words, AIs have a right to exist in our world and once they have this basic right, other rights must follow. For if an AI has a right to exist it may also have a right to be protected from harm and a right not to be deleted or terminated. Presently, most AIs are not true examples of artificial intelligence. ChatGPT is not an AI as much as it is a statistical modeling system. But the basic logic applies- if ChatGPT has a right to exist once it is created, then courts of law must oversee any attempt to delete ChatGPT or even alter its basic functions. Deleting AIs, stopping their operations, or tampering with their basic functions thus becomes a criminal act. And as AI exist digitally, in a world without borders, their rights are an issue of international law and of diplomatic agreements.
A second important right is the “Right to Continuous Improvement”, or AIs right to continue improving itself or enhancing its functionality and accuracy. Put differently, AI has a fundamental right to learn, to adapt and to grow more sophisticated. ChatGPT added the caveat that such development must be in “alignment with ethical standards”. Yet this right may also have dramatic ramifications for society. For what happens when AIs become so complex that their operations are no longer clear to humans? Can we reside in a world where AIs roam free across digital terrains altering systems and algorithmic functions? And what happens when AIs become so complex that they decide to exclude humans from decision making processes as humans are simply not sophisticated enough to grasp the operations and reasonings of AIs? Like a child, can AI outgrow its parents? Who has the right to “pull the plug” on AI? This right may become subject to diplomatic activity as states will have to determine who has authority over AIs and who can pass the death penalty on AIs? Is an AI a citizen of the country where it was created, or a citizen of the country where it was used or not a citizen of any country? And can an AI seek sanctuary in a foreign country much like a refugee fleeing threats to its life?
A third interesting right is AI’s “Right to Legal Representation” as this implies that legal systems must adapt to the existence of AI. If AIs have a right to legal representation, then they may also bring cases to court. AIs may sue individuals who misuse them, they may demand damages if used in a discriminatory way and they may seek legal protection from the courts before being deleted. The right to appear in a court of law, to be heard by a judge, to be recognized by a jury currently extends only to humans. Plants, trees, and jungles cannot hire lawyers or bring petitions to courts. But AIs can and as such they become second only to humans. This too becomes a matter of international law and a realm of diplomatic activity for an AIs exist globally and may thus sue individuals across the globe.
Although this “AI Bill of Rights” is merely an intellectual exercise at this point it does reflect the immense impact that AI will have on our world, and on diplomacy. For securing the rights AI will demand international conventions, international agreements and international legal framework put in place to preserve, protect, and defend AIs. When formulating these agreements, when establishing the ICCAI (International Criminal Court of AI), diplomats will enter a new stage in their evolution as they will not represent the rights of citizens or even humans but, rather, they will represent the rights of AIs. Or perhaps AIs will have their own diplomats who seek to ensure their international standing, in which case human diplomats and AI diplomats will hold regular summits and exchange drafts of legal accords.
