February 28, 2017
Opinion: Robots and AI could soon have feelings, hopes and rights … we must prepare for the reckoning
Get used to hearing a lot more about artificial intelligence. Even if you discount the utopian and dystopian hyperbole, the 21st century will broadly be defined not just by advancements in artificial intelligence, robotics, computing and cognitive neuroscience, but how we manage them. For some, the question of whether or not the human race will live to see a 22nd century turns upon this latter consideration. While forecasting the imminence of an AI-centric future remains a matter of intense debate, we will need to come to terms with it. For now, there are many more questions than answers.
It is clear, however, that the European Parliament is making inroads towards taking an AI-centric future seriously. Last month, in a 17-2 vote, the parliament's legal affairs committee voted to to begin drafting a set of regulations to govern the development and use of artificial intelligence and robotics. Included in this draft proposal is preliminary guidance on what it calls "electronic personhood" that would ensure corresponding rights and obligations for the most sophisticated AI. This is a start, but nothing more than that.
If you caught any of the debate on the issue of "electronic" or "robot" personhood, you probably understand how murky the issues are, and how visceral reactions to it can be. If you have not caught any of it, now is a good time to start paying attention.
The idea of robot personhood is similar to the concept of corporate personhood that allows companies to take part in legal cases as both claimant and respondent – that is, to sue and be sued. The report identifies a number of areas for potential oversight, such as the formation of a European agency for AI and robotics, a legal definition of "smart autonomous robots", a registration system for the most advanced ones, and a mandatory insurance scheme for companies to cover damage and harm caused by robots.
The report also addresses the possibility that both AI and robotics will play a central role in catalysing massive job losses and calls for a "serious" assessment of the feasibility of universal basic income as a strategy to minimise the economic effects of mass automation of entire economic sectors.
As daunting as these challenges are – and they are certainly not made any more palatable given the increasingly woeful state of geopolitics – lawmakers, politicians and courts are only beginning to skim the surface of what sort of problems, and indeed opportunities, artificial intelligence and robotics pose. Yes, driverless cars are problematic, but only in a world where traditional cars exist. Get them off the road, and a city, state, nation, or continent populated exclusively by driverless cars is essentially a really, really elaborate railway signalling network.
I cannot here critique the feasibility of things such as general artificial intelligence, or even the Pandora's Box that is Whole Brain Emulation – whereby an artificial, software-based copy of a human brain is made that functions and behaves identically to the biological one. So let's just assume their technical feasibility and imagine a world where both bespoke sentient robots and robotic versions of ourselves imbued with perfect digital copies of our brains go to work and "Netflix and chill" with us.
It goes without saying that the very notion of making separate, transferable, editable copies of human beings embodied in robotic form poses both conceptual and practical legal challenges. For instance, basic principles of contract law would need to be updated to accommodate contracts where one of the parties existed as a digital copy of a biological human.
Would a contract in Jane Smith's name, for example, apply to both the biological Jane Smith and her copy? On what basis should it, or should it not? The same question would also need to be asked in regard to marriages, parentage, economic and property rights, and so forth. If a "robot" copy was actually an embodied version of a biological consciousness that had all the same experiences, feelings, hopes, dreams, frailties and fears as their originator, on what basis would we deny that copy rights if we referred to existing human rights regimes? This sounds like absurdity, but it is nonetheless an absurdity that may soon be reality, and that means we cannot afford to laugh it off or overlook it.
There is also the question of what fundamental rights a copy of a biological original should have. For example, how should democratic votes be allocated when copying people's identities into artificial bodies or machines becomes so cheap that an extreme form of "ballot box stuffing" – by making identical copies of the same voter – becomes a real possibility?
Should each copy be afforded their own vote, or a fractional portion determined by the number of copies that exist of a given person? If a robot is the property of its "owner" should they have any greater moral claim to a vote than say, your cat? Would rights be transferable to back-up copies in the event of the biological original's death? What about when copying becomes so cheap, quick, and efficient that entire voter bases could be created at the whim of deep-pocketed political candidates, each with their own moral claim to a democratic vote?
How do you feel about a voter base comprised of one million robotic copies of Milo Yiannopolous? Remember all that discussion in the US about phantom voter fraud, well, imagine that on steroids. What sort of democratic interests would non-biological persons have given that they would likely not be susceptible to ageing, infirmity, or death? Good luck sleeping tonight.
These are incredibly fascinating things to speculate on and will certainly lead to major social, legal, political, economic and philosophical changes should they become live issues. But it is because they are increasingly likely to be live issues that we should begin thinking more deeply about AI and robotics than just driverless cars and jobs. If you take any liberal human rights regime at face value, you're almost certainly led to the conclusion that, yes, sophisticated AIs should be granted human rights if we take a strict interpretation of the conceptual and philosophical foundations on which they rest.
Why then is it so hard to accept this conclusion? What is it about it that makes so many feel uneasy, uncomfortable or threatened? Humans have enjoyed an exclusive claim to biological intelligence, and we use ourselves as the benchmark against which all other intelligence should be judged. At one level, people feel uneasy about the idea of robotic personhood because granting rights to non-biological persons means that we as humans would become a whole lot less special.
Indeed, our most deeply ingrained religious and philosophical traditions revolve around the very idea that we are in fact beautiful and unique snowflakes imbued with the spark of life and abilities that allow us to transcend other species. That's understandable, even if you could find any number of ways to take issue with it.
At another level, the idea of robot personhood – particularly as it relates to the example of voting – makes us uneasy because it leads us to question the resilience and applicability of our most sacrosanct values. This is particularly true in a time of "fake news", "alternative facts", and the gradual erosion of the once proud edifice of the liberal democratic state. With each new advancement in AI and robotics, we are brought closer to a reckoning not just with ourselves, but over whether our laws, legal concepts, and the historical, cultural, social and economic foundations on which they are premised are truly suited to addressing the world as it will be, not as it once was.
The choices and actions we take today in relation to AI and robotics have path-dependent implications for what we can choose to do tomorrow. It is incumbent upon all of us to engage with what is going on, to understand its implications and to begin to reflect on whether efforts such as the European Parliament's are nothing more than pouring new wine into old wine skins. There is no science of futurology, but we can better see the future and understand where we might end up in it by focusing more intently on the present and the decisions we have made as society when it comes to technology.
When you do that, you realise we as a society have made no real democratic decisions about technology, we have more or less been forced to accept that certain things enter our world and that we must learn to harness their benefits or get left behind and, of course, deal with their fallout. Perhaps the first step, then, is not to take laws and policy proposals as the jumping-off point for how to "deal" with AI, but instead start thinking more about correcting the democratic deficit as to whether we as a society, or indeed a planet, really want to inherit the future Silicon Valley and others want for us.