Moral Turing Test

The principle of the Moral Turing Test is that only qualifed Moral Agents can identify other Moral Agents. This principle ties in with other ethical ideas I have put forward; viz., that moral agency is self-declared and that moral injunctions evaporate. While several readers find themselves uncomfortable with all of it, they should not. Each of these ideas is perfectly reasonable and comports well with many other commonly accepted values. When determining competency, for example, it is a person or group which decides; it is not a simple matter of grading a test made up of "objective" questions.

Moral evaporation with respect to maxims of behavior amounts to Kantian 'similar circumstances.' Moral rules are only imminent when the situation warrants. Whether one uses drugs, has sex, or does anything else is entirely circumstantial, even if maxims can be formulated in the universal form 'Anyone in this situation should do this.'

On the proverbial desert island, whatever one does or does not is entirely left to its sole resident. This, it seems to me, is the zero order case, or the central fact. In the absence of any society, each person decides whatever is good or right or none of the above. There is neither standard nor consequence beyond the individual's beliefs and actions. If the stranded person wishes to attribute all that happens to some greater forces, that is just as plausible as attributing everything to a whim of imagination. There is no debate about it, and no proof is required, because meaningful debate and proof are social phenomena; i.e., such activities require an "other," not self.

That each of us declares oneself a moral agent is just to say that each of us originally lives on a desert island. That this is so, regardless of the real or imagined status of others, is clear from the fact that our thoughts, feelings and perceptions are confined in our heads. Despite the supposed "oneness" people achieve through sex and other activities, there is observably, always more than one persona involved. "Oneness" is a temporary feeling, not a fact. It is only our individuality that is the constant fact, and the underlying reason why moral agency is always a matter of self assertion.

From this starting point, the first order case is that ethics (and almost everything else) is a social activity. A strong case can be made, for example, that even science is a social activity, not some "objective" body of knowledge. Kuhn's theory of paradigms depends on the claim that scientists are conformists who cling to received teachings. Like every other tribe, they resist new ideas until defense of the old ways becomes untenable. Well into the 1950s, I heard people question Einstein's Relativity which, by that time, was considered "proven" by most scientists. Those latter day objections of the few had been widespread among scientists in the early 20^th century. Unlike science, ethics and other cultural matters are not decided by designated officials at field tests. Each of us has an interpretation which is believed and carried out until death or social intervention, which makes it a lot harder to establish common ground. But the lack of common agreement, or even a procedure for arriving at common agreement, does not in itself obviate the social nature of ethics.

The Moral Turing Test, or Moral Agent Test, is inherently social. The Moral Agent Test is modelled on the Turing Test for intelligence, hence my naming it the 'Moral Turing Test.' (I think the credit for an idea should go where it is due.) Here is my operational description of how it works. In this test, there are one or more judges who decide whether those submitted to examination qualify as moral agents. To make the test suitably "objective" and "scientific," the judges will not be informed  about those tested except to the extent the judges demand information and those tested give it. Again, we attempt to place what is tested in a black box. The judges only see what is emitted from the black box, not what is inside. The judges are free to impose any test they determine in their sole discretion on the black boxes. That discretion includes potentially destructive tests, so it is possible the judgement will be self-defeating.

The Moral Agent Test is a generalization of the Turing Test. There may be one or more judges. There may be one or more examinees. Their roles may be reversed; in fact, all particpants may be both judge and judged at the same time. Each particpant may reach different conclusions about any or all of the other participants. Sometimes participants may act in concert (socially) to reach a consensus opinion. Whatever conclusions they reach, only those who are moral agents by self-declaration are capable of determining whether others are also moral agents. This situation is analogous to John Rawls' original position, except that we make no assumptions about "reason" or "rationality." This definition is purely operational.

Now, the purpose of this arrangement is that the judges will have to act ethically in the process of making of their determination. This is not to predetermine what is ethical. In fact, in the process, the judges will be explicating what they consider ethical. So, the judges are inherently part of the determination; i.e., the process is recursive. The notion of recursion is the simplest one I can put forward in analogy to the process, because recursion involves calculation of a result based on existing, previous result(s). This is a way of getting around the conundrum presented by Bertrand Russell about sets within sets, or the sets of all sets. (This followed from Whitehead & Russell, Principia Mathematica.) We can define the process, whether or not it has a finite length or limit, and let it run. It is sufficient that we can build the next iteration. This is the same sort of thinking that leads to Mandelbrot's fractals and other methods of analyzing chaotic systems. What we have is the process, potentially never ending or potentially self-destructive or potentially inconclusive. We do not know which in advance; we get there by going there. In this sense, the Moral Agent Test is also Darwinian.

Now, taking apart the test a little, it is important to note that the judges must be capable of moral judgement in order to judge the competence of anyone else. We use this criterion in appointing Judges to our Courts, and teachers to our schools. While several tests are applied to those appointed - nominees are vetted - the presumption is that those doing the vetting are capable of picking good judges or teachers. The tests are merely an adjunct to the making of a judgement; usually, a social necessity to assure observers that makers of the judgement have similar thoughts and share common ground. This leads to an infinite regress, so how can they do that? Some people just "look" judge-like, teacher-like or whatever-like; it's a feeling. A modern realization is that our judgements are fallible, so we almost always subject those appointed to a trial period (probation) before granting tenure. This is a sensible solution to the uncertainty of good judgement when the actual standard is probably unknown. In effect, we do an experiment and then qualify the outcome by whatever criteria we have.

One of the major differences of modern times from ancient times is the acceptance of uncertainty. We are building it into our societies at every level. Not everyone can accept this; uncertainty is upsetting. It is the flight from uncertainty that fuels religious fundamentalism, religion generally and modern Conservatism. Since the Moral Agent Test is built on uncertainty, it is a modern idea which cannot be religious or conservative. The Moral Agent Test is like democracy: it subjects a problems to individual judgements and accepts the outcome, at least tentatively.

The Moral Agent Test is not reciprocal. That is, those being tested, if they are Moral Agents, may or may not have the same opinion of their judges. This is not an unreasonable outcome, as judges and the judged often disagree completely about the verdict, especially if they are of different racial or ethnic groups. Criminals often protest their innocence, even while the Guillotine blade drops. Minorities often allege racism as the motivating factor of their conviction or abuse, Meanwhile, those on the other side of the bar see nothing wrong with what has transpired. Lynching, for example, was an unexceptional event for many White people in the Old South, but had traumatic effects among the black population. To say that the Moral Agent Test is not reciprocal is thus to say that values (judgements) are relative as well as uncertain.

In the case of meeting ET (have we already met it in the form of our animal relatives?), our judgement about it or them need not be reciprocal. Meeting an ET has most of the necessary prerequisites for our Moral Agent Test. We have no idea what's "inside" the alien any more than ET apprehends us. We start tabula rasa. Are moral rules species speciifc? Are there any general ethical principles? How do we behave in comparison to our ET?

One of the easiest examples of how species and cultures differ on supposed "ethical principles" concerns sex. Various TV channels present programs about so-called primitive peoples, many of whom live in tropical areas and wear few clothes. The relative nakedness of those viewed seems to be a criterion of "primitiveness" and titillating for most European and Asian audiences. Yet, the same nearly bare rear ends and breasts are easily visible on most First World beaches during the summer. Who is primitive? When?

The sorts of ethical principles I propose should be the generalized results of moral agency and judgements. But they are a slippery bunch, as can be easily seen by examining differing sexual mores alleged to be moral matters. I resolve any seeming contradictions by putting sexual mores in the same class as food practices. Almost every culture has developed unique table manners and food rituals which can be as extraordinarily complex, subtle and time consuming as the Japanese Tea Ceremony or as brutally sloppy as American fast food joints.

There's no judging the results of moral agency in advance. Things can turn out in wholly unexpected ways. This makes dealing with others always uncertain, if they are anything but zombies or robots. This presents an opportunity to those who offer simple, pat solutions to problems, because many, maybe most, people cannot tolerate confusing, erratic results. The pat answer comes in many forms. In the moral world, it is Ten Commandments, Papal Bulls and all the other authorities which must be obeyed absolutely and accurately: 'Just do as you're told, and you will be saved.' In my experience, most people like that; they prefer not to think about how to lead their lives. But I believe that is false; to lead such a life is to lack "authenticity." (Here, I must take on an existentialist vocabulary.) It is not even possible to take the results of several judgements as given, as indicating; every situation is potentially contrary. In principle, scientific truths have the same status.

But it would be exhausting to rexamine every decision, every minute. For this reason, we develop theories about the decisions we have made. Then we allow those theories to guide us. This is an acceptable procedure, provided we do it with our eyes wide open. There always has to be testing, the possibility of falsification.  The art or science of moral agency is Ethics. That is, ethics is the theoretical structure we raise over the specific instances of moral decisions. Ethics can only be invented by those sufficiently intelligent to recognize themselves as moral agents. Thus, there is a connection between intelligence and morality. That connection, whatever it is, also links the Turing Test for intellgience and the Moral Turing Test for moral agency. All of these tests are recursive.

I think all of these things developed naturally through Darwinian evolution. That is, there was a progression from our early ancestors, who did not recognize themselves as moral agents, to ourselves who do. I am sorry to frustrate those who want clear cut distinctions, but I doubt very much there was any First Moral Agent. Perhaps there was an early Homo ? who wondered, 'should I do this?' That wondering, a conscious uncertainty, may have been the beginning of our modern assertions of the voluntary and the ought. But I think it unlikely we will ever know who, exactly, it was that first wondered. Moreover, even were we present on that occasion, I think different people would have different opinions about the nature of that wondering. It would have only been much later, when speech or signs were well developed, that general agreement would probably be reached, certifying the creature as a moral agent.

How did we become moral agents? We pulled ourselves up by our bootstraps. Are we moral agents? Well, I know I am one.