Experience Machine Ethics

The following post was originally written as a paper for a class I took at Northwestern University this past spring.

 

One of the most enduring themes in science fiction has been that of an experience machine: a system that gives its users immersive, customized experiences on demand. From Star Trek to The Matrix to X-Men to Doctor Who, viewers and readers have for decades struggled with the implications of the immense, potentially double-edged power such machines would wield. Since the dawn of the 21st century, however, such dialogues have taken on a new urgency, fueled by the rise of sophisticated digital systems that, while by no means capable of creating entirely immersive customized experiences, have begun to be used in ways which would suggest them to be a sort of proto-experience machine: consider recent developments in virtual reality, gaming, and customizable digital entertainment, for instance. Therefore, to give thought to the philosophical, ethical, and practical ramifications of the genesis of experience machines at this point in time is no longer analogous to speculating about the implications of interstellar human travel — it is probably more analogous to mid-20th century philosophers considering the potential implications of a future humanity connected to a universal communications network.

This paper will seek to examine the ethical problems that would be presented by the ultimate emergence of a complete experience machine, and subsequently present a code of ethics intended to guide institutional responses to these ethical problems. As we will see, the experience machine will ultimately represent not just a mind-boggling new technology, but a critical milestone in the macro-scale development of humanity.

 

One of the first commentators to seriously consider the implications of the genesis of an experience machine was Robert Nozick, in a brief portion of his 1974 book Anarchy, State, and Utopia. Nozick immediately points out in this book perhaps the most obvious ethical dilemma that would be created by experience machines: people might very well have no reason to leave the experience machine. After all, why should one waste one’s time living in the real world (where many things are difficult and unpleasant) when one could simply live in a world that feels real in which things are guaranteed to be easy and pleasant? It is perhaps likely that even Nozick understates the sheer magnitude of this potential problem. Much of the first several paragraphs of Nozick’s treatment of the experience machine concern whether one should plug in to it. Ample evidence, however, shows that humans reliably and voraciously take any opportunity they can to escape reality: the average American, for instance, escapes from reality for over five hours per day, by watching television in its various forms. And this is simply escape from reality using a medium as relatively primitive as a moving picture screen — imagine how much more seductive a complete experience machine would be. Ultimately, the ethical problem of people would never leave the machine leads to a variety of sub-problems, too: why reproduce with a human (a very complicated affair) when you can simply have the experience of sex in a machine on demand? Why work, or do any of the real-world things that can be rewarding, but demand effort? These questions are largely practical ones, but they can easily become ethical issues, too, when a utilitarian moral lens is applied to them; issues that coming generations of humans will almost certainly need to grapple with.

Another potential ethical problem associated with the experience machine is the way it might affect how humans think. If one spends all day in a pliable world built to bring one happiness, how will that person begin to view the real world differently? In the real world, for instance, one often must interact with difficult individuals, a class of person that would not exist in the experience machine. It is not unforeseeable that people used to interacting with experience machine “people” would have no patience whatsoever for difficult real-world individuals, lacking the (important) skills necessary for conflict resolution, mediation, et cetera. More insidiously, someone who spent a lot of time in the experience machine would almost certainly be used to sex on demand, without any way of navigating the complicated (but incredibly important) topics like consent that surround real-world sex. A humanity plugged into the experience machine en masse might therefore be a humanity full of sexual assaulters with nonexistent people skills.

The experience machine might also generate perverse financial incentives. In the modern world, the primary method of disseminating technological advances has been (and will almost assuredly continue to be) capital markets. Depending on the market structure used to deliver experiences to the end user (e.g., if experience machines are themselves sold, or if time in them is sold), the companies selling experiences might very well have an incentive to keep the user in the machine as much as possible — potentially at the cost of the user’s physical/emotional health. It is not hard to imagine an experience machine in which experiences are designed to keep the user using the machine, much like today’s social media platforms are designed to keep the user clicking, scrolling, and liking. Just as Facebook has deployed advanced behavioral psychology to make its product as addicting as possible, so might experience machine companies make experiences as addicting as possible (and imagine how much more addicting an experience machine might be than a social networking website!) The only foreseeable solution to this perverse incentive structure would be selling experience machines themselves directly to the end user, so that revenue for the experience machine company is not exactly directly correlated with its product’s addictiveness. However, depending on how expensive each experience machine is, this business model might not make sense.

A final ethical problem likely to arise from the genesis of experience machines is their potential usage for unpleasant experiences. One can easily imagine experience machines replacing waterboarding as the world’s interrogation technique of choice. The possibilities along these lines are incredibly frightening: imagine being trapped in a nightmare both fully real and from which one cannot possibly wake up. Not much more need be said about this possibility, but it is worth bearing in mind that experience machines could be used as the ultimate torture as well as the ultimate entertainment.

 

To guide mass usage of experience machines, a code of ethics is in order. The following one is largely motivated by the aforementioned ethical problems associated with experience machines, and seeks to balance consideration of what is right with what is feasible. It is primarily addressed to corporate makers of experience machines and government regulators.

 

The Mostly Comprehensive Experience Machine Code of Ethics

 

1. No entity — government, corporate, or otherwise — should seek to impose a limit on the amount of time individuals can use the experience machine or seek to alter the way in which the user desires to use the experience machine, with the following exceptions.

1. 1. Interference is permissible if the user displays signs of experience machine-induced mental illness. Every experience machine should come with a built-in evaluator of the emotional/psychological/mental health of its users that uses machine learning algorithms to detect behavior known to be anomalous. If the evaluator raises any red flags, the user should be gently returned to the real world, and designated for psychiatric help.

1. 1. Experiences that are very likely to make their experiences dangerous to others in the real world may be pre-censored, so that no user may obtain that experience. Such experiences may include those likely to alter the user’s underlying psychology in a detrimental way.
2. Full cognizance of the power of the experience machine should be incorporated into the machine’s design and operating norms.
   1. Participation in an experience machine should require express written consent.
   2. Participants should never be held in an experience machine against their will.
   3. Every experience machine should have a “kill switch” that, when activated, immediately returns the user to the real world.
   4. Participants should have full veto control over whatever experiences they may be having. While machine-generated experiences are permissible, they should always be able to be overridden by the user.
3. Experience machines should be operated with awareness of the financial incentives they may generate.
   1. Experiences should be entirely anonymous. Under no circumstances should data from the experience machine be mined and re-sold to advertisers, or provided to government agencies.
   2. The utmost efforts should be taken to promote individual ownership of experience machines, potentially even in the form of a government subsidy.

 

While some aspects of this code of ethics may seem self-explanatory, others are likely to require extensive normative justification and descriptive explanation. What thus follows is a clause-by-clause discussion of the values represented in the code of ethics as well as how those clauses are likely to interface with the ethical problems laid out at the beginning of this paper.

 

There can be said to be two core values underlying the Mostly Comprehensive Experience Machine Code of Ethics. The first is techno-liberalism: that is, an enduring commitment to liberal values in light of new technological developments. The second is anti-parochialism: an unwillingness to view technological developments outside of their historical context (in the case of the experience machine, that of the history of entertainment, very broadly.) These core values are elucidated upon in the upcoming clause explanations to which they are most relevant.

The first clause of the Mostly Comprehensive Experience Machine Code of Ethics is largely founded in the idea that the right of human individuals to make choices should be respected, in accordance with fundamental liberal values. It thus fails to delineate more than the two specific cases in which oversight of individuals’ usage of the experience machine is permissible. While it may seem that this first clause sets a dangerously high bar for interference in such a manner (to the point that it would likely not actually prevent the first ethical problem — over-usage — from occurring), it should be noted that to take a more conservative approach to the experience machine’s usage is to take a strong stance on the moral value of entertainment in general. Any argument that might be given in favor of restricting usage of the experience machine (such as the fact that it wastes time, isn’t productive, etc.) could also be applied towards restricting usage of Netflix, for example. It is important to note that the experience machine would be the terminal innovation in entertainment, not an entirely new class of entertainment — this is the core insight of anti-parochialism as applied to the experience machine. That is, the experience machine would unite and improve upon many different aspects of entertainment that humans currently consume — including television, video games, and pornography — and yet those current iterations are distributed more-or-less uninhibited (largely due to contemporary society’s adoption of liberal values.) The class of ethical problems at which the first clause is aimed, therefore, are likely to be unsolved in the era of the experience machine.

The second clause of the Code of Ethics is largely motivated simply by concern for the well-being of experience machine users, as well as by techno-liberalism. With regards the likelihood of implementation: subclauses a., c., and d. seem very likely to be implemented naturally by market incentives, whereas, especially if the technology becomes widespread, subclause b. seems rather unlikely to be adhered to. As long as there are governmental regimes with a need for extracting information from prisoners, the experience machine seems likely to be used as the ultimate enhanced interrogation technique, unfortunately. Nonetheless, there is undoubtedly value in making a normative statement on the subject. This code of ethics’ prohibition of experience machine torture again rests in techno-liberalism: respect for the human rights of detainees, even in the era of experience machines. Hence, the subclause’s inclusion, despite the relative unlikeliness of its implementation.

Finally, the third clause of the Mostly Comprehensive Experience Machine Code of Ethics is motivated directly by a techno-liberal commitment to preserving the sanctity of human minds — thus, the clause’s efforts to combat addiction financially. If experience machines are too expensive for most people to afford, but cheap enough that most people could afford to rent time in them, their individual purchase might be subsidized by the government (for qualified buyers, at least.) Some might balk at this proposition, given its nontraditional use of public funds. However, this likely understates the extent to which the experience machine would be the ubiquitous terminal innovation in entertainment. The ability to have customized, seamless entertainment on demand might be so compelling that access to it might begin to be seen as a fundamental human right — in which case its subsidization might well be seen as perfectly sensical. Thus, subsidization of access to the experience machine would kill two birds with one stone: it would ensure that the experiences are not made unnecessarily addicting, and that most people would have equal access to it.

 

It is hardly deniable that the above code of ethics, if implemented, would do little to hamper usage of the experience machine, potentially even usage en masse. What if, one might argue, virtually everyone plugged into the experience machine, and was never heard from again in the base reality? Wouldn’t that cause problems both ethical — billions of people would live in a “false” reality — and practical — no one would be around to feed everyone?

Firstly, it must be pointed out that the potential practical problems associated with this possibility are ultimately inferior to the ethical ones, since it is naïve to think that the free markets would not supply a method of (for instance) supplying the plugged-in humans with nutrients, given the scale of economic demand that would exist for such a service. The ultimate question is thus whether a scenario almost akin to a self-imposed version of The Matrix (in which billions of humans are permanently in a near-vegetative state while plugged into the experience machine) is ethically desirable.

To reach such a conclusion, it is helpful to once again invoke anti-parochialism. As mentioned in an earlier section of this paper, the average Westerner dedicates a very significant amount of time to entertainment — an amount that our not-so-distant ancestors (in addition to citizens of third-world countries) would find obscene. Since the experience machine will really just be the consummate version of the entertainment we now consume, those of us in 2018 might do well to remove the proverbial log from our own eye before attending to the speck in our distant successors’. That is, given our contemporary culture of entertainment — we spend some 33% of our waking hours consuming it, after all2 — one might begin to suspect that detractors wouldn’t really have an ethical objection to plugging into the experience machine (even for long periods of time) if they had the option to.

Secondly, if every human plugs into the experience machine on a quasi-permanent basis, they would spend their lives living in a world explicitly designed to bring them happiness. Since the experience machine could (in principle) emulate the base reality to the smallest iota, one would have no reason to believe it could do any worse at bringing happiness to the user than the base reality. Therefore, since it is a commonly held principle that one should generally respect others’ pursuit of happiness (it is, after all, enshrined as a human right in such terms in the United States’ Declaration of Independence), who is to say that their neighbor has no right to pursue happiness by permanently plugging into an experience machine?

Finally, to assign superior ethical value to the base reality is to fundamentally misunderstand some of its attributes. In order to make the argument that a virtual reality has less inherent value than the base one, one must recourse to denigrating it on the basis that it is only made “real” in the user’s mind. But the same can easily be said of the base reality! One can interpret Descartes’ famous observation cogito, ergo sum as consisting precisely of the recognition that reality is only real insofar as we make it so. In the Critique of Pure Reason, Kant likewise observed that “all appearances are together to be regarded as mere representations and not things in themselves, and accordingly that time and space are only sensible forms of our intuition, but not determinations given for themselves or conditions of objects as things in themselves. To this [idea] is opposed transcendental realism, which regards space and time as something given in themselves.” If one were to spend one’s life, therefore, from birth in an experience machine identical to reality, then removed from it in middle age and forced to spend the rest of one’s life in the base reality, the results would be no different than if the realities had switched places. More concisely, both the base reality and a virtual one would be based entirely on axioms — admittedly, axioms of varying obviousness — but all axioms nonetheless. It is therefore extremely difficult to maintain in a philosophically consistent manner the position that it is unethical for the world’s humans to plug into the experience machine en masse.

 

Upon its genesis, the experience machine is likely to be a technology that introduces new ethical problems, compounds old ones, and forces a reconsideration of many aspects of what it means to be human. A code of ethics, therefore, that puts the experience machine in its appropriate historical context and keeps in mind what so many humans hold to be the most important liberal values is of paramount importance in bringing this new technology to a stable, safe, and powerful fruition.

Nonetheless, it is perhaps somewhat paradoxical to speak of putting the experience machine in historical context, for while the ways in which humans use it may have historical context, the technology itself most certainly will not. It has been put forward as the terminal innovation in entertainment — but entertainment itself may prove to have an expiration date sometime in the not-too-distant future. As we continue to discover the intricacies of the human brain, and as digital systems continue to grow in sophistication and scale, the two seem likely to join in an ever closer union, to the point where no distinction between man and machine can truthfully be drawn. It is not unlikely that those future humans, capable of editing their brain-function at will, would have little use for something as pointless as entertainment. Therefore, the experience machine may represent the final technological invention of the pre-cyborg era, and thus should be viewed in the most serious of lights. It might be said that as humanity gradually assembles the parts for a complete experience machine, it will simultaneously assemble a monument to the twilight of its primitive era.